❌

Vue lecture

Apple’s OS release and how it impacts Identity Graphs

Back in September, Apple rolled out its yearly update to the Mac OS, iPad OS and iOS operating systems.  Since then, advertisers making use of identity resolution services such as ID5, The Trade Desk’s adsrvr.com endpoint and LiveRamp (amongst others) have noticed those requests begin to fail to resolve – impacting the ability for the advertising site to leverage those services to enrich their identity graph which affects targeted advertising efforts downstream.

It turns out this was an intentional change introduced by the Webkit team, and they’ve come out with a new explainer discussing it available on the Webkit blog.  The change will affect MacOS, iPad OS and iOS 27 and higher.  

All Roads Lead to Networking

This feature will use a dynamically generated list from a remote service.  When a webpage attempts to call a 3rd party domain a series of checks occur, with validation against the dynamic list.  If a match comes up that this domain is used for identity resolution for advertising – the system will terminate the network connection prior to any data exchange.  

On the webpage – the JavaScript console will throw an error with the message “Blocked by tracking protections” with the script path.

This change occurs at the network level, under the web browser.  Due to the nature of the checks involved changing the parameters on the tag won’t allow it to resolve. This impact to advertising is not easily addressable and it is recommended to adjust marketing plans rather than attempt to win a game of cat and mouse by seeking to outsmart the network classifier.

This change will impact both Safari and Webkit (and thus every browser that runs on iOS and iPadOS).  While adoption of the new release is still early, historically adoption surges in the 4th quarter and enters the 80-90% mark shortly after the new year.  If this trend holds – impact for affected services will increase during the 4th Quarter retail push, potentially resulting in less sales due to weakened ability to effectively apply targeted advertising to marketing plans. 

Interactions with Marketing Tech

If the marketing stack makes significant use of identity resolution services to power marketing efforts it’ll be important to note that the total addressable audience will decrease as the OS update adoption increases.

This in particular can result in breakage in cross-domain and cross-device customer tracking – which is by design.  This can/will have downstream effects on marketing efforts based on those audiences.  

For scenarios which handle conversion tracking,  that too will be impacted when reliant on identity services.  If the system can’t generate an identity (due to the network request being blocked) this can change how the data that appears in reporting.  This may manifest in yet more results being classified as ‘Direct” as no stitching of the prior user behavior would have occurred to drive attribution rules when the visit crosses domain boundaries.

Interactions with Optimization Tech

It’s not customary for A/B testing platforms to make use of external identity resolution services, so while you can and should check for impact I would not expect most systems to suffer degradation of performance due to this change.

However if you are using 3rd party identity resolution to drive personalization efforts – those are very likely to be impacted and thus periodic testing is advised.

Interactions with Privacy Tech

It’s worth noting that the change to block the network domains will occur regardless of the user consenting to the Consent Banner.  The user may say “I Agree”, but the technology will block the data transfer by default.  Users can disable the feature, but as it’s enabled by default – I suspect the number of users who will disable this will be in the low single digits.

Privacy teams should be aware of this so that they can have productive discussions with Marketing about how it’s not the consent banner at fault.

Is Server Side the Answer?

Not necessarily.  Many Conversion APIs will have a client side component which does the initial identity establishment.  That identity is then used to power the Server to Server Conversion API. These types of services will be adversely impacted because the initial identity assignment will fail. This means that any call that then relies on that identity won’t have it – and so many behave in an unexpected / undesired manner.

It’s also not overly simple from a networking point of view.  Safari already looks at CNAME redirects when resolving network traffic.  Attempts to cloak this traffic will be discovered by the network resolver and impacted accordingly.

So could server to server work?  Yes.  However it requires refactoring how data flows between systems, which means it’s likely to be time intensive / costly.  As many retailers enter a code freeze in early Q4 due to the holidays – this doesn’t leave much, if any, time to refactor marketing data flows in advance of the final push leading up to Christmas.

Instead of panicking, I would recommend reviewing the impact on your marTech stack and work with developers & vendors to determine how to best rebuild how the website moves data around. Of course, even if you do move to a server to server connection, data privacy laws continue to apply, and if you are now running middleware to shift to first party identity resolution / data transfer – that comes with all the attendant maintenance, compliance and security overhead. This should be strongly considered prior to making any investment in refactoring.

The post Apple’s OS release and how it impacts Identity Graphs appeared first on Privacy & Security.

  •  

AI as Infrastructure

I recently attended the IAPP Global Privacy Summit and noticed something I have seen repeated in various podcasts.  The conversation around Artificial Intelligence is changing. No longer are we asking if we should use AI, or having discussions on how to use AI, instead we’re talking about building on top of AI powered systems.  In this context, AI and the related systems it runs on are infrastructure supporting other business functions.

This shift in discussions is notable and means how we think about AI has to change.  If AI truly is infrastructure, then I see a disconnect between how people approach using AI to support business processes, and actually treating it as infrastructure backed by sound engineering considerations. So in this post I want to highlight just some of the things I think deserve serious consideration when deciding to build a business / feature on top of an AI system. 

Contracts and Compliance

Many companies are subject to various legal and industry concerns that they have to consider when making business decisions.  Often contracts are used to handle various aspects such as who owns what, and what controls may need to be in place.  This is particularly important when dealing with standards such as SOC II or laws such as Europe’s General Data Protection Regulation.  

Typically a business would not actively seek to deploy systems which jeopardize these considerations. However, which service and what license level employees may use can put compliance with those standards and regulations at risk. 

Alex Isskova created a handy chart that helps to illustrate this issue:

Generally speaking, free accounts mean that the service can train on the data you are feeding it.  This could range from a business risk to outright illegal under various conditions.  It’s important when treating AI as infrastructure you consider what it truly means to build a service offering on top of an AI platform. Different contracts may be required (such as Enterprise agreements), different documentation may have to be written, various risk assessments may have to be done and chances are those items often have to happen prior to actually launching the system for public use.

Business Continuity 

The next thing I have noticed relates to business continuity.  Everyone wants to talk about what happens when the AI system works as expected.  Very few people I have spoken with have had serious discussions about what happens if the AI system fails or becomes otherwise inaccessible.  In technology downtime isn’t a matter of if, it’s a matter of when.  When you have critical business systems that need to be running – consideration needs to be given to failure cases.

An example may be helpful.   Let’s say you’re using Google Cloud Platform and leveraging Gemini in your project.  Your project is set up exclusively in a data center on the East Coast.  What happens when the data center goes offline (due to a storm or some other condition)?

The wrong time to find out about this is when it actually happens.  The correct time is now.  If the AI is infrastructure supporting other business processes  and you’re subject to up-time requirements then typically work is required to ensure that the AI and related systems can actually meet those Service Levels.   

High Availability of critical systems is based on good engineering practices and considering the failure cases in detail. The wrong time to do this is when the system is offline and already impacting business processes.  If AI truly is infrastructure – then you need to be proactive in ensuring the system can meet business objectives.

Security

The last thing I want to touch on is often the conversations I see mention agentic automation of specific tasks.  With this process the user is attempting to automate some activity they regularly undertake and hand it off to an AI process to conduct. In many cases the user leverages their own accounts to conduct this automation.  There are security considerations with this.  Two of the major ones relate to access levels and a concept called non-repudiation. 

When it comes to access levels, most users are an administrator on their personal computers.  This means if the AI is granted the same access as you have – it can effectively do everything you can do.  This means that it can open a file, run spell check and send an email, but it can also reformat your hard drive.  The danger here can be extensive considering that some AI agents have been shown to execute commands from simply viewing a webpage at the same permission level it was running at.

AI systems need to be properly scoped with permissions to limit the damage they can do.  This typically means provisioning the AI system with a low access level service account because generally you’d want the agent to fail to conduct a task rather than risk it destroying a system. 

Allowing the AI to use the same user account as the person behind the keyboard poses another risk, one related to the security concept of non-repudiation.  In non-repudiation, you ensure that a party in a transaction cannot deny the authenticity of their message or actions. This breaks down when multiple people or systems share accounts. This isn’t just a matter of ensuring people can’t deny actions, it’s critical for good governance of administrator accounts. If something goes wrong you want to be able to determine if it was a user mistake or an AI making the error.  If everything looks like a user in the logs, it can be difficult to identify the source and address it in a timely manner.

Conclusion

It can be exciting to build things with AI.  However serious discussions have to happen when converting that AI powered system into infrastructure for your business.  As soon as you start building on top of an AI system several risks surface that can imperil business operations if not addressed.  I think it’s neat that people are thinking about AI as infrastructure, but if it truly is infrastructure – then we need to treat it like it is, and that means taking the time to ensure it’s fit for the business purpose we’re matching it to.

The post AI as Infrastructure appeared first on Privacy & Security.

  •  

Risky Business In the New Year

Are you a marketer?  Perhaps an analytics or data science professional?  Maybe an Optimization lead?  Then this blog is for you.  The below processes are very likely to impact you due to the regulatory requirements.

This past September California finalized risk assessment requirements for the California Consumer Privacy Act (CCPA). These regulations are very prescriptive as to when an assessment is needed, what goes into a risk assessment, and when reporting on these assessments to the State needs to occur.

Note: The following talks about regulations and risk assessments. You are advised to seek out qualified counsel or risk personnel to assist you with your specific situation.

The risk requirements begin on page 83 of the finalized regulations and will require that companies subject to the CCPA attest to the California Privacy Protection Agency (CPPA) by:

By April 1, 2028, a business covered by the CCPA must submit to the CPPA:

  1. An attestation that required risk assessments were completed, and
  2. A summary of their risk assessment information.

Compliance with the risk requirements begins January 1st, 2026 with the first mandatory reporting occurring in April of 2028. For assessments occurring for 2027 and later, attestation of completion occurs in the following year on April 1st.

While the initial timeline seems generous – given the complexity of the assessments and the scope of impacted systems, affected companies are strongly encouraged to prioritize compliance activities in the short term. 

Separate from the attestation – The CPPA or the California Attorney General may request copies of risk assessment reports at any time, and these must be provided with-in 30 calendar days. This adds increased importance to conducting these assessments in the identified manner.

When is an assessment required?

Subsection a of Section 7150 states that a risk assessment is required whenever data processing presents significant risk to the consumer’s privacy. Notability, subsection b goes on to state in the first condition that selling or sharing personal information presents significant risk.

Under the CCPA – the definition of personal information is very broad.  Personal data includes any data that identifies, relates to, or could reasonably be linked to a person or their household – either directly or indirectly.  This includes (but is not limited to):

  • Name
  • Email Address
  • Purchase History
  • Browsing History
  • Location Data
  • IP Address
  • Profiles of the consumer
  • Sensitive personal information

This means that – the following activities will likely trigger this type of assessment:

  • Targeted Advertising (both targeting and measurement)
  • Most kinds of analytics both software and analysis.
  • Most kinds of A/B testing both software and analysis
  • Most website 3rd party “widgets”.  Such widgets very likely get the user’s IP Address.

Relevant Considerations for Agencies / Contractors:

  • You’ll need to do assessments for your own website / mobile app and related services
  • You may be expected to assist your clients with the completion of their own assessments, in particular for any service or product you recommend. 

Given the sheer scope of affected systems and processes, I expect these assessments to take  a non-trivial amount of time. This can affect launch timelines, as these assessments are required to be done prior to the data collection / processing. 

It is reasonable to think, “this will slow me down” and yes it likely will – and that’s by intent. Section 7154 outlines the goals of a risk assessment:

The goal of a risk assessment is restricting or prohibiting the processing of personal information if the risks to privacy of the consumer outweigh the benefits resulting from processing to the consumer, the business, other stakeholders, and the public.

Stakeholder Involvement

When faced with new regulations, a lot of teams seek to offload the work to Legal / Compliance groups.  The regulations account for this (Section 7151)  – and mandate that business employees support the assessment process by making information available for the assessment.  

For example – if Marketing wants to collect more personal information for advertising, the specific employees who made that determination will need to be part of the assessment process.

This means that the business and engineering groups will likely need to be part of any assessment conducted on behalf of the business, which is likely to be a change in process.

What’s in an assessment?

An assessment is a complex analysis that looks at risk to the consumer from multiple angles.  In addition to the stakeholder involvement, outside sources may be required to complete specific parts of the assessment.

Risk Assessment Requirements – identify and document:

  • The business purpose for processing consumer personal information.  This can’t be in generic terms such as “improving services”.  If a topic like “improving services” is used – it needs to be specific in how those services are improved.
  • The categories of personal information to be processed, inclusive of any sensitive personal information.
  • The businesses planned methods for collecting, using, disclosing, retaining or otherwise processing personal information, and the sources of that personal information.
  • How long the business plans to retain each category of personal information, or the criteria the business plans to use to determine that retention period.
  • The business’s method of interacting with the consumers whose personal information the business plans to process, and the purpose of that interaction.
  • The approximate number of consumers whose personal information the business plans to process.
  • What disclosures the business has made or will plan to make to the consumer about processing their information.
  • The names or categories of the service providers, contractors or third parties that personal data will be disclosed to, and the purpose for which it is disclosed.
  • The benefits to the business, the consumer, other stakeholders and the public from the processing of personal information. Note, this can’t be in generic terms.
  • The negative impacts to the consumer’s privacy associated with the processing.  The business must identify the sources and causes of these negative impacts.
  • Economic Harms, including limiting or depriving consumers of economic opportunities.
  • Physical harms to the consumer or property, including processing that creates opportunity for physical or sexual violence
  • Reputational Harms, inclusive of emotional distress, stress, anxiety, embarrassment, fear, frustration, shame and feelings of violation that could negatively impact an average consumer.  The regulations give examples of disclosure of nonconsensual intimate imagery or purchase of pregnancy tests may trigger emotional distress from disclosure.
  • Any safeguards that the business plans to implement for the processing to address the negative impacts identified above.  Examples include security, encryption, use of privacy-enhancing technologies or the consulting of external parties.
  • if the business will initiate the processing post risk assessment.
  • all individuals who provided information for the assessment, except for legal counsel who provided legal advice
  • The date the assessment was reviewed and approved, and the names and positions of the individuals who reviewed and approved the assessment, except for legal counsel who provided legal advice. The regulations specifically call out that a person who has the authority to participate in deciding whether the business will conduct the processing covered by the assessment must review and approve the assessment.

If the assessment covers Automated Decision Making Technology (ADMT)  – additional information needs to be part of the assessment:

  • The logic of the ADMT, including any assumptions or limitations of the logic; and
  • The output of the ADMT, and how the business will use the output to make a significant decision.

As you can see, this process touches on multiple areas and may include several team members to conduct.  It is unlikely that these reports can be generated quickly if done as intended.  It is advised to build in enough time up front – because these assessments must be completed prior to conducting the activity covered in the assessment for any activity begun after January 1st, 2026.

Assessment Timing and Retention

A business must comply with several conditions around when to conduct and update an assessment:

  • A business must conduct and document a risk assessment in accordance with the regulations before initiating any processing activity identified in Section 7150 – subsection b.  For most cases – this will be the selling or sharing of personal information.
  • At least once every three years – the business must review, and update if necessary the risk assessment to ensure they remain accurate.
  • A business must update the assessment as soon as possible, but with-in 45 days – of a material change relating to the processing activity.
  • The business must retain the risk assessments, including both original and updated versions, for as long as the processing continues, or for five years after the processing completes, whichever is later.

For activities that started prior to January 1st 2026, and continue after January 1st 2026 – the business has until December 31st 2027 to conduct the risk assessment.  Any new activity after January 1st 2026 needs an assessment prior to conducting the activity.

Attestation to the Agency

Beginning April 1st 2028 and every year thereafter a business must attest to the State of California that these assessments are being conducted.

A business must submit to the California Privacy Protection Agency:

  1. The businesses name, and point of contact for the business, inclusive of name, phone number and email address
  2. The time period covered by the submission, by month and year.
  3. The number of risk assessments conducted or updated by the business during the time period for the submission, both in total and for each of the processing activities identified.
  4. Whether the risk assessments conducted or updated during the time period of the submission involved the processing of personal information and sensitive information . 
  5. Attestation under penalty of perjury.  Perjury in California can include fines of up to $10,000 or up to four years in prison.
  6. The name and business title of the person submitting the risk assessment information and the date.

The person submitting must be a member of the businesses executive management team who:

  • Is directly responsible for the business’s risk-assessment compliance
  • Has sufficient knowledge of the businesses’s risk assessment to provide accurate information; and
  • Has the authority to submit the risk assessment information to the Agency.

In closing

As you can see – these assessments are extensive and can cover a wide array of information.  They will mandate inclusion of business stakeholders and involve Executive attestation.  These processes may be new for businesses unfamiliar with data privacy laws, and discussions may need to happen with the Executive team so they understand this new reality.

The turn of the year will prove interesting and if the above is anything to go by – expect new processing activities to take longer and require more paperwork.  With the CCPA Security Requirements also looming, multiple business teams will have to find new ways of working as the calendar changes.  

The best advice I can offer is to be proactive about it.  These processes may consume a large amount of time.  It’s best to begin work on these activities and have the required executive discussions in the short term so you can best prepare for if the California Attorney General comes knocking that you will have something to show them when prompted.

The post Risky Business In the New Year appeared first on Privacy & Security.

  •  

Security Requirements come to the CCPA

It’s October, Cybersecurity Awareness month, and I thought this would be a good talk about the recently finalized cybersecurity requirements issued by the California Privacy Protection Agency(CPPA) for the California Consumer Privacy Act(CCPA). These regulations are the first time a State level privacy law has been very descriptive on what they expect and I suspect it will serve as a blueprint for other States in the future.

Note: The following talks about regulations and security. You are advised to seek out qualified counsel or security personnel to assist you with your specific situation.

The cybersecurity requirements begin on page 88 of the finalized regulations and will require that companies subject to the CCPA attest to the CPPA by:

  • April 1, 2028, if the business makes over $100 million;
  • April 1, 2029, if the business makes between $50 million and $100 million; or
  • April 1, 2030, if the business makes less than $50 million.

So while the initial glance at these deadlines may cause businesses to think they have time prior to enforcement, it’s worth noting the scope of the assessments before drawing such a conclusion. Depending on the current state of a cybersecurity program, there may be considerable work to be undertaken in order to pass an audit. That work effectively cuts down the time that may be available to begin preparing.

What’s in an assessment?

The regulations make clear that if a audit is required, it must assess how the business established, implemented and maintains it’s security program. The auditor (discussed below) must evaluate specific things as defined by regulations, as well as anything else they feel applicable to the business.

The audit must assess – if applicable:

  • Authentication – and makes note of Multi-Factor Authentication and specific password requirements.
  • Encryption – both at rest and in transit
  • Account Management and Access Controls
  • Inventory and Management of personal information used by the business’s information system
  • Secure configuration of hardware and software (including items such as patching, cloud, and change management)
  • Internal and External vulnerability scans, and penetration testing.
  • Audit-Log Management
  • Networking monitoring and defenses, including items like Data Loss Prevention systems, bot detection intrusion detection and intrusion presentation.
  • Anti-Virus / Malware protections
  • Segmentation of the system (firewalls, routers, switches)
  • Limitation and control of ports, services and protocols
  • Cybersecurity awareness (how the business maintains knowledge of current cyber threats and countermeasures)
  • Cybersecurity education and training (inclusive of contractors)
  • Secure Development and coding best practices, inclusive of code reviews and testing.
  • Oversight of service providers and contractors (Third Party Risk Management)
  • Data Retention Schedules and proper disposal of personal information.
  • How the business manages responses to security incidents (Incident Response Plans)
  • Business-continuity and disaster recover plans.

Depending on the current state of a cyber program it can easily take some organizations years to design and implement all of the above. It is strongly encouraged to begin work early to ensure you have enough time.

The Audit Report

California has two pages of requirements for information on what the report must contain. Notably, if the business undertakes other audits (such as SOC 2 Type II or ISO 27001) it may be able to reuse those, provided it also meets all the requirements under the CCPA regulations.

Certification

The business will be required to conduct an audit and submit certification to the agency no later than April 1st of the following year the business was required to complete the audit.

Notably however, this attestation must be completed by a member of the business’s executive team that is responsible for cybersecurity audit compliance, has sufficient knowledge of the audit to provide accurate information and has the authority to submit to the agency. The executive would have to submit the information under penalty of perjury. This is a shift in risk and liability that should be discussed with the executive team well in advance of the mandatory submission deadlines.

Choosing an Auditor

California also has requirements over who can be considered an qualified auditor. It will be critical for businesses to ensure their auditors have the proper qualifications prior to undertaking an audit.

Auditors are required to be qualified, objective, and independent. They must use the procedures and standards of the profession of auditing such as those from one of the following organizations:

  • American Institute of Certified Public Accountants (AICPA)
  • The Public Company Accountability Oversight Board (PCAOB)
  • The Information Systems Audit and Control Association (ISACA), or
  • International Organization for Standardization (ISO)

It should be noted several of the above organizations have mandatory work requirements to qualify as a certified auditor. ISACA for instance has the Certified Information Systems Auditor certification. This certification has a 5 year work experience requirement, but waivers for education in the field can reduce this. The important thing to realize here is not just anyone can qualify for auditor certification. It typically requires work experience and business will need to verify this while selecting an auditor.

They additionally need knowledge of cybersecurity and how to audit a businesses cyber program. While the regulations do not go into much detail here, I would venture this would be best reflected in degree work and / or professional certifications in cybersecurity.

The regulations do state however, that the auditor can be internal. However, if this is the case the auditor must be independent. Effectively the team responsible for security can’t check their own homework or take direction from management.

The auditor will evaluate the program based on the specific evidence provided by the business, but can not rely primarily on the assertations of the businesses management. This means that the team responsible for compliance will need to have everything documented for the auditor. This documentation process can consume large amounts of time and so I encourage affected companies to begin work in preparation sooner rather than later.

The post Security Requirements come to the CCPA appeared first on Privacy & Security.

  •  

AI Regulation is Here

Since the end of 2023 with the release of ChatGPT, the discussion around artificial intelligence(AI) has exploded. While initially this caught many regulators off guard, they have been rapidly introducing regulation in effort to avoid what many perceptive as the falling to regulate digital advertising.

Where We Are Now

For most organizations dealing with the United States and Europe, a few notable laws have already passed and begun enforcement. This includes Utah’s Artificial Intelligence Policy Act (on May 1,2024), Colorado’s Artificial Intelligence Act (which will enter enforcement Feb, 1, 2026) and the European Union’s colossal Artificial Intelligence Act (which has multiple phase in dates over the next several years).

We also recently saw amendments be passed, such as this amendment in California which extends the protections of the California Consumer Protection Act to AI systems, when that AI system deals with personal information of California residents.

It thus be reasoned that adding personal information for jurisdictions that have privacy laws may bring those laws into scope in addition to any specific AI laws that also apply. Organizations are advised to really understand their use case and the data involved to help determine regulatory obligations to avoid being blindsided with unexcepted enforcement actions.

What Is Coming

It could be said these laws are precursors for that is to come. As of earlier this month State legislatures have already introduced more bills than in all of 2024.

just 66 days into 2025 and the number of artificial intelligence bills pending in the US now officially exceeds the total for ALL of 2024. 👀

2024: 743 AI bills total for the year
2025: 781 AI bills in just 66 days pic.twitter.com/xrYPVDYgR1

— Adam Thierer (@AdamThierer) March 7, 2025

It is exceptionally unlikely that out of these hundreds of bills, that nothing will pass and become law. Still, there are common themes between what has already been enacted, and what is proposed. Forward thinking organizations would be well served from designing and implementing an AI Governance Program – something which many of the passed regulation and proposed bills require.

AI Governance Programs

A common misconception is that all the risk sits in the AI Model itself. A AI Risk Management framework helps to illustrate the required / suggested processes for managing the AI system (inclusive of it’s hardware, relevant software, Modal, Security and Privacy concerns).

To this end, several organizations have released an AI Risk Management Model that may be worth considering:

These frameworks seek to introduce controls that enable or support items such as:

  • Explainability
  • Accountability
  • Transparency
  • Fairness
  • Robustness
  • Saftey
  • Security
  • Data Governance; and
  • Privacy

Often the AI laws require extensive documentation of the AI risk analysis (which can often be requested by regulators), and several have specific data retention requirements on this documentation . Organizations would be well served by factoring in the time investment for these activities in any AI effort.

AI Security

In addition to the above governance program, I want to drill down a bit into the security aspect. Many of these regulations mandate security of the AI system. AI Security is what you do in addition to having a robust security program already in place as it builds on it. Several laws / bills mandate external validation for specific use cases, to be conducted by an experienced Penetration Team. This means that security is not something that can be quickly bolted on and needs to be considered early in the design process as it can have fundamental impacts on the overall system architecture. Organization are advised to have trained Cyber Security personnel involved in the AI system design and deployment to assist in these matters.

Thinktanks such as OWASP have come up with a top 10 list for various AI related concerns that is worth reviewing. It should be realized that while the top 10 are easily listed, they may require sizable effort to implement mitigations depending on the use case.

Summary

AI regulation will continue to take center stage for the next several years as States domestically and countries internationally seek to put laws on the books to clarify liability and reduce risk to the general populace. Any organization undertaking AI work is thus strongly advised to consider proactively implementing an AI Risk Management program, and ensuring that proper trained professionals are involved in the design and implementation any AI system. These are actions that can be taken now, to avoid pain later once enforcement picks up in the coming months / years.

The post AI Regulation is Here appeared first on Privacy & Security.

  •  

The FTC and Hashing

Earlier last week the Federal Trade Commission published a blog about how the hashing of data does not, in fact, make the data anonymous. This was news to many people who do not deal with security or privacy law regularly and that’s understandable because few people take the time to really develop a strong understanding of cryptographic concepts1. It’s made more complex by the fact that the legal definition of anonymous and what most people consider anonymous are not the same.

The Legal Definition

Laws vary – but here in the United States most companies attempt to align to California’s Consumer Privacy Act. That law doesn’t make reference to anonymous data, but does to ‘de-identified’ data as reads below:

(m) “Deidentified” means information that cannot reasonably be used to infer information about, or otherwise be linked to, a particular consumer provided that the business that possesses the information:

(1) Takes reasonable measures to ensure that the information cannot be associated
with a consumer or household.

(2) Publicly commits to maintain and use the information in deidentified form and not
to attempt to reidentify the information, except that the business may attempt to
reidentify the information solely for the purpose of determining whether its
deidentification processes satisfy the requirements of this subdivision.

(3) Contractually obligates any recipients of the information to comply with all
provisions of this subdivision

https://cppa.ca.gov/regulations/pdf/cppa_act.pdf

This is the type of definition the Federal Trade Commission is talking about. The law is clear in that the information can’t be used to infer information about a specific consumer. It’s in this way that the law conflicts with the realities of hashing. Hold on to this understanding, because we’ll loop back around to it.

The Technical Reality of Hashing

At a high level, a hashing algorithm is a one way function that, for a specific input, returns a specific message digest. So provided you have the same input, and know the specific hashing algorithm used – you should be able to create the same digest. The process is deterministic in this regard and while it does obscure the source data (the hash is one way) it’s not anonymous. When specific algorithms are found to have collisions (in that two inputs can return the same digest, the algorithm typically gets retired for a more secure replacement.

Some examples of common use cases:

Validating Files

You can use a specific hashing algorithm, such as SHA2, and input a specific piece of data (such as a software executable file) and create a hash. That hash can then be compared to the hash published on the software publisher’s website to determine if the software is genuine or has been modified since it was published, and is therefore untrusted.

Digital Signatures

Digital signatures are used to verify the authenticity of digital messages and documents. This relies on public key asymmetric cryptography to help provide assurance that both that the message is real, and non-repudiable (they can’t claim they didn’t see it / sign it). If you’ve ever signed a contract online – this process is how that becomes legally binding.

Marketing Attribution

Lately, many vendors such as Facebook have been requesting companies provide hashed information for attribution purposes (in Facebook’s case, for the Conversion API). Facebook requests that the information be transmitted with SHA256, but as you can see from the documentation for each specific input, once ran through SHA256, you end up with an expected output. In this way the data is obscured, but not anonymous because it is linkable back to the specific input.

Problems with Hashing

Hashing has a few specific problems in the context of anonymization – but let’s look at one of the most common.

Rainbow Tables

A rainbow table is a table of precomputed hash values. that acts as a lookup index. Since the hash of a given input is deterministic you can precompute the entire key-space and run a basic SQL query where a where clause of the digest to return the original value used to provide that digest. In this way the cryptography isn’t broken, but because the hashes match identification of the original source is possible. If that source matches a user – then the user is identifiable. With the use of cloud services on demand, building a rainbow table and storing it (they are rather large) is simply a matter of being willing to pay for the compute and storage.

Defense against such attacks is possible via the use of cryptographic salts and peppers, but as you can see from the Facebook documentation referenced earlier these are not used for the Conversion API because Facebook wants to attempt to match the information to their own records to ensure attribution for Facebook Ads. Google does the same thing for their advertising platform with their Enhanced Conversions feature.

Issues with True Anonymization

Legally, true anonymity is a high bar to reach and may be impractical for specific use cases.

You anonymize a dataset by transforming it in key ways, often with statistical techniques such as k-anonymity, l-diversity and injecting noise with t-closeness. This process is very much like the way the Census Bureau in the United States seeks to protect those who take part in the census. Typically the more anonymized a data set becomes, the harder it becomes to obtain insight or derive value from it.

Various attacks do exist to reconstruct the data set in whole or part based on information left in the set so ensuring a data set remains anonymous is a moving target. It’s important to remember that just because you can’t reverse engineer a data set, does not mean that it’s impossible. Also noteworthy is if you can reverse a data set, than it’s not anonymous and becomes subject to different legal requirements.

The Legal Issue

Now that we have a grounding in hashing, and we know it’s deterministic and understand how a hashed digest’s original value can be determined – we see we can’t meet the threshold laid out in the CCPA for deidentified data when it comes to the marketing use case for attribution because for attribution the entire thing is working on the premise of being able to reidentify the user through a lookup table on the ad publishers platform in order to determine return on ad spend and related metrics.

This does mean however, as that the data may instead qualify for the definition of pseudonymization used by various laws. Typically data that is pseudonymized is still required to be disclosed to the user upon request and still subject to technical and organizational measures to protect such data. Under Europe’s General Data Protection Regulation – you are still required to obtain consent prior to collection. This also may impact the specific contractual elements of a data processing agreements or privacy notices if you were referring to hashed data as anonymous.

1: If you want to develop a in depth understanding of Cryptography, I recommend this book.

The post The FTC and Hashing appeared first on Privacy & Security.

  •  

The CCPA Standard

As this year began, there were five comprehensive state privacy laws in effect.  However, by the year of 2026, this will more than triple with, at the time of writing, 17 States having passed similar laws. Against this backdrop, and with a lack of Federal Standard, I have seen a tendency to adhere to the ‘strictest’ law, or the State with the largest economy, which would be California.

From a practical standpoint this makes a lot of sense.  It strives to place a company in the best position for additional laws as they come into effect and helps to streamline operational processes and thus increase efficiency.  However, there is danger in assuming that California’s standard is the only one you’d have to adhere to.  Below are two key differences organizations should consider as they strive to operate in this new environment.

The below is not legal advice – please consult legal counsel for determining how anything may apply to your specific scenario

Not all risk assessments are created equal

While the California Consumer Privacy Act (CCPA) calls for the need of risk assessments the California Privacy Protection Agency has yet to fully define what exactly that looks like in the regulations.  Other States, such as Colorado have been a bit more proactive.

Colorado’s regulations (Part 8) define the need for a data protection assessment, and are much more specific to what needs to be considered (coming in at roughly two and a half pages of requirements).  It’s unlikely a data privacy professional can answer all of the various considerations exclusively under their own power.  Colorado seeks evaluating conditions such as – psychological harm, financial harm, security harms and so on.   Some of these may require a very specialized level of knowledge to adequately evaluate.  Partnerships will be key in ensuring a valid evaluation can be conducted.

It may be that California’s requirements, once issued, are more stringent than that of Colorado, however until that time Colorado in my opinion has one of the most in-depth evaluation processes required for a data protection / risk assessment.  Standardizing on the California requirements may not meet the standard of Colorado, but Colorado, by being more complete, may satisfy the eventual requirements of California and other states.  This is just one example of where a different State law may provide a better baseline for compliance activity.

NonProfit Status doesn’t always mean exempt

While most states (such as California) do exempt nonprofits Colorado, Delaware, Indiana, New Jersey and Oregon cover nonprofits in their data privacy laws.  This means that even as a nonprofit, you may still have obligations under one or more of those laws.   Should your nonprofit operate in, or target users of those States you may have work to do in order to ensure compliance with the respective laws.  This is particularly true if you have never undertaken this work before for other compliance reasons.    

Further, every State now requires contractual terms when enlisting service providers / data processors.  It may be that you become subject to the laws (even in States other than the above) depending on who your clients are and what they may  subject the nonprofit to by contract.  Consideration should be given to how these new laws come into effect impact your nonprofit, and proper funding / staffing should be obtained to ensure compliance obligations are met, ideally ahead of enforcement deadlines. 

Conclusion 

Data Privacy laws will continue to take center stage for the next several years.  As these laws are refined there will increasingly be a need to standardize to a baseline.  In the United States, that baseline should likely be California.  However, as shown above, there are scenarios where other States may require additional compliance work.  It would be dangerous to assume that just because you are compliant with the CCPA that you could pass compliance obligations for a regulator in a different state or country. 

In all likelihood it will get more difficult before it gets easier.  Unless Congress passes a pre-emptive Federal law, I fully expect the patchwork of laws to continue to grow, and the requirements to comply with all of them to continue to expand.  Organizations will be well served to ensure they are periodically reviewing their compliance programs, and adapting them as each new law goes into effect.  Privacy compliance is a moving target, and until that changes you’ll never be done. This is why it’s so important for anyone involved in the field (or who funds those teams) to understand the current reality. The proper level of investment is required from the company / organization so it can better respond to changes as new laws enter enforcement.

The post The CCPA Standard appeared first on Privacy & Security.

  •  

Blog Update

Greetings!  

As you may have noticed, over the past year my cadence of posts has decreased.   There has been a good reason for this, and in this post I plan to inform you what I have been up to, and what to expect going forward.  However, to understand what is coming and why I did what I did, a look at some history may be in order.

A Focus on Data Protection

Prior to Europe’s General Data Protection Regulation (GDPR) entering enforcement, I was responsible for tagging marketing/analytics pixels in my role at the time.  As a result of this, I was already keeping track of the various technical changes (such as Safari’s Intelligent Tracking Prevention) that were impacting data collection / data quality.  

When the direction came down we had to become ready for GDPR. I was assigned to the project as the engineering resource due to my familiarity with the data collection that was in place. I decided to assist the lawyers and read the entire GDPR text prior to my first meeting with them.  This turned out to be a key factor in how fast we could move and I became a sort of translator between the broader technical team and the lawyers.   I greatly enjoyed this interaction of law, technology and security, and volunteered to repeat the process the following year as we made ready to comply with California’s Consumer Privacy Act (CCPA).

In late 2018, I started blogging and writing about various privacy, security and legal topics due to my work on these projects. For the rest of my time at that employer, I would be involved in privacy discussions, as well as assist our computer security team with various training exercises, interfacing with the legal team as new concerns arose.

A Desire for More Formal Training

I would, in 2020, eventually switch companies to Further, where I currently work.  As part of my day to day here, I am able to leverage my previous experience to assist our clients with data privacy questions.  This ranges from explaining various technologies under consideration due to privacy concerns to assisting with consent management and data governance work as well as educational training for clients on related topics.

As the regulatory and technical environment became more complex, I sought to further my training in privacy engineering by completing IAPP’s CIPT and CIPP United States & CIPP Europe exams.   This helped close some gaps I had in my knowledge, particularly in newer technology applications, but towards the middle of 2022 I started wondering what I could do next to continue to enhance my knowledge of data protection. 

The Path I Selected

I started looking around at different certifications and programs, and would come to realize I had an issue that would make the next step difficult.  In the early 2010s, I graduated college with an Associate of Science, in Information Systems Management, which had a concentration in Web Application Development.   To this point I had done, I feel, remarkably well with the AS, but a lot of the courses I had interest in were a Masters or equivalent level.  Faced with this, I ultimately opted to enroll in a Bachelors program in order to be able to later unlock the coursework I wanted, transferring in the credits from my Associate’s degree.

I wanted to stay in the realm of data protection, and since dedicated privacy courses don’t really exist at the bachelors level, I started looking at pre-law, and cyber security as possible options.  After a lot of reflection, and quite a bit of research into both options.  I ran an analysis of trends in recent data privacy laws and regulations. I felt a grounding in cyber security would serve me better as an undergrad and enable me to leverage my decade of experience in software development.   

Cyber Security may seem a odd choice. Commonly, privacy work is seen as an intersection of security and law.  From my view, privacy can exist without law (encryption goes back hundreds of years), but not without security. Further, nearly every new data protection law has the phrase “adequate technical, organizational and administrative safeguards” when it comes to the collection and storage of personal information, but seldom do the laws or their regulations actually define what “good” looks like.  I thought this was interesting and would be a good course of study to drill into as a result.

With this decision made, I applied and was accepted to a school and this brings us to the last several months where I have been doing coursework on a range of cyber security topics.  I also attempted and completed several certifications from CompTIA and ISC2 to validate I had sufficient understanding of the material I was learning.

In February 2024, I completed my final courses and earned a Bachelors of Science in Cyber Security and Information Assurance.  This continues my focus on data protection and opens a few doors for continued education that were previously closed to me, while also enhancing my ability to review technical designs for compliance obligations from more than one lens. 

What’s Next

Foremost, my goal is to get back to posting on this blog more regularly for the remainder of this year.  The content will continue to focus on privacy topics, with perhaps a bit of an increase in security related material where it makes sense.

Secondarily, I plan to complete a Masters program in Cyber Security and earn a few more certifications that I would like to acquire in the security field.   That said, the Masters program I selected is not as demanding as the undergrad curriculum I completed and so I feel I will have more time to support posting in the future through the rest of the year.

At this point, I will either be done with formal education, or seek to enroll in law school in a part-time program.  Previously this wasn’t an option for me without relocation and earning a BS degree.  However, I now have a BS degree and the American Bar Association has since begun to approve distance learning programs placing this option on the table.  Should I opt to pursue this I would endeavor to tailor the education around contract law, cyber security and data privacy / protection topics to whatever extent is possible.  I would be very curious to those lawyers who read this blog, your thoughts on the matter and any advice you may have to provide on this possible path.

So in closing, I have a rough path of where I want to go, and what it may take to get there – but I wanted to let you, the reader, know I am not done posting about Privacy and Security topics, and I expect to pick up cadence in the coming weeks.

The post Blog Update appeared first on Privacy & Security.

  •  

2024 Comes in Like a Lion

Mere hours ago, we bid as we were bidding farewell to the year 2023, Utah’s data privacy law become enforceable. This action will be a single item across a chain of events that will force companies to adapt to the new world as we move into Q1 2024. Certainly this new year will force a re-evaluation of various key activities that take place in the online MarTech ecosystem. In this blog post, I highlight some of the key changes taking effect in the first quarter.

The 3rd Party Cookie Phase Out Begins

Next up will be Google’s phase out of 3rd party cookies, beginning with 1% of Chrome Users (roughly 30 million users) on January 4th. Google is encouraging organizations to adopt their privacy sandbox initiatives. The impact from this change will show in analytics and performance monitoring and may lead to some uncomfortable conversations with leadership when graphs and charts appear to be going in the wrong direction. Education activities for stakeholders are strongly advised to avoid very uncomfortable questions at reporting time.

It should be noted that while the phase out starts at 1%, it will rapidly increase and is expected to complete sometime in the 3rd quarter. Organizations which have failed to account for this shift in technology may find some website features and key marketing activities such as remarketing stop working. This effectively means if you were relying on such technology you must come up with a replacement process. If remarketing accounted for most of your marketing budget, a re-evaluation may be in order.

Apple Will Begin to Enforce Privacy Manifests

For those who have mobile applications on Apple’s AppStore – it should be noticed that Apple will begin automatic enforcement of privacy manifest files. They have also released a list of third party SDK requirements. Apps which use these SDKs have additional actions to take in app development as well as when submitting the app via App Store Connect.

Once automated enforcement begins apps will be blocked at submission that are missing the required disclosures / manifest files. Organizations operating on the AppStore are running low on time to ensure their software is compatible with the new requirements ahead of the Spring 2024 enforcement.

Google Gets Serious on Consent Mode

For those organizations operating in the European Economic Area (EEA) they should be aware of the Digital Markets Act. As part of this act, various ad publishers will impose new requirements to ensure they are compliant with the law. The Consent Made changes are how Google is addressing compliance and if you want to handle measurement and advertising in the EEA Google will require the Consent Mode signal to be in place prior to March 2024.

Google is very clear that in order to continue from benefiting from measurement, ad personalization and remarketing features that consent must be obtained (as is required by the ePrivacy Directive and GDPR). This includes data sent to services from Google Analytics and even extends to offline data such as offline conversion reports.

If you use MarTech services in the EU – you must complete the requirements prior to March, or you will begin to suffer data loss. This should not be a major lift for organizations already compliant with the ePrivacy Directive and GDPR, but it still requires explicit action on the part of the organization. This work may need to go through change management in some organizations so should be penciled in for work at the earliest opportunity to avoid disruption to operational efforts.

New Health Data Laws

The following is not legal advice and you should consult with counsel if you believe it may impact you.

Beginning on March 31st, 2024, the States of Nevada and Washington will begin enforcing their respective health privacy laws. These laws extend in scope beyond what is covered in HIPAA (Health Insurance Portability and Accountability Act) and may impact businesses not traditionally considered healthcare. Here’s an example of the data covered under Washington’s My Health My Data Act:

“Consumer health data” means personal information that is linked or reasonably linkable to a consumer and that identifies the consumer’s past, present, or future physical or mental health status.

For the purposes of this definition, physical or mental health status includes, but is not limited to:

(i) Individual health conditions, treatment, diseases, or
diagnosis;
(ii) Social, psychological, behavioral, and medical
interventions;
(iii) Health-related surgeries or procedures;
(iv) Use or purchase of prescribed medication;
(v) Bodily functions, vital signs, symptoms, or measurements of
the information described in this subsection (8)(b);
(vi) Diagnoses or diagnostic testing, treatment, or medication;
(vii) Gender-affirming care information;
(viii) Reproductive or sexual health information;
(ix) Biometric data;
(x) Genetic data;
(xi) Precise location information that could reasonably indicate
a consumer’s attempt to acquire or receive health services or
supplies;
(xii) Data that identifies a consumer seeking health care
services; or
(xiii) Any information that a regulated entity or a small
business, or their respective processor, processes to associate or
identify a consumer with the data described in (b)(i) through (xii)
of this subsection that is derived or extrapolated from non-health
information (such as proxy, derivative, inferred, or emergent data by
any means, including algorithms or machine learning).

Nevada’s law is very similar to that of Washington, but with a key difference. Nevada’s law does not allow for private action, where Washington’s does. This dramatically increases the risk for organizations which are believed to be in violation Washington’s law may be subject to lawsuits from the general public in addition to the Attorney General for Washington.

The Washington Attorney General has released a F.A.Q. around the law that may be worth a review.

It is strongly recommended that organizations discuss these laws with their legal counsel and take any required corrective action before enforcement begins at the end of March.

The post 2024 Comes in Like a Lion appeared first on Privacy & Security.

  •  

New Firefox Privacy Enhancements

Last week Firefox released version 120, just in time for Black Friday. As part of the update Firefox released a number of new privacy enhancements, some of which will impact various analytics, marketing and compliance efforts.

Global Privacy Control Support

Firefox has had support for the Global Privacy Control (GPC) signal for some time, but it was locked behind configuration settings and wasn’t available in the main UI settings. With release 120 this has changed and support for the GPC signal can now be found under Privacy & Security. Mozilla details new change and what it means in a knowledgebase article and press release.

The GPC signal has long been considered to be a mandatory opt-out under California law, and with additional states considering the signal for their own opt-out mechanics (which we’ll find out more about in 2024). Sites would be well served to add in support for the GPC Signal in order to avoid possible enforcement actions by regulators.

It should be noted that the exact expected behavior when detecting a GPC signal may vary by State, and steps should be taken to ensure that proper handling occurs for the State(s) you may operate in. Generally however, this means that data stops being transferred to third parties (as legally defined) and will result in data loss for the affected systems.

Anti-Fingerprinting Technology Enhancements

Firefox will now apply Fingerprinting Protection to the HTML Canvas APIs when ran in Private Mode or ETP-Strict Mode. In effect this will cause the browser to add in noise to the Canvas APIs to reduce the ability for Canvas features to be used as a fingerprinting mechanic. Identity solutions which rely on fingerprinting should be evaluated to determine what impact this change may have.

Note that Firefox fights fingerprinting (all browsers do) and that fingerprinting a device may not be legal in your jurisdiction. Caution is advised.

Privacy Tests in Germany

Two major privacy changes are being tested in the German market.

Cookie Banner Blocking

Seeking to limit the annoyance of cookie banners, Firefox will now attempt two actions when running in Private Mode per the knowledgebase article.

First, the browser will attempt to inject a cookie on as part of the page load, which will be set to the ‘declined’ state. This should prevent the banner from opening, as it will believe you have already made a consent selection.

In cases where this isn’t possible, Firefox will instead search for and click the banner’s “decline” or “reject” button. This would dismiss the banner as if you are interacted with it yourself.

If you’re interested in how this works, you can check out the Github repo.

Consent Management Solutions impacted by the feature include TrustArc, OneTrust, Cookiebot, Didomi and others.

Sites impacted include some of the most popular sites in Germany, and a full list of vendors and sites can be seen here.

URL Query Parameter Stripping

The first of these, URL Tracking Protection by default in private windows – will remove query parameters that are often used to track users. This is testing promotion of a feature which is currently only enabled under the Strict security setting to be the default behavior in Private Mode.

With both of these features, while Germany is the initial market, based on performance Mozilla plans to roll out the features to additional areas.

The post New Firefox Privacy Enhancements appeared first on Privacy & Security.

  •  

FTC Warns Over Improper Data Collection

On September 18th, 2023 the Federal Trade Commission (FTC) sent a letter to five tax preparation companies over possible unfair and deceptive practices. These letters state that data collection for marketing and advertising purposes, when gathered in a confidential context, could be a violation that could subject the company to a fine. An overview of this action can be found in the press release.

We can see from the Recipients of the Notice that this includes:

  • H&R Block
  • Intuit
  • TaxAct
  • TaxSlayer
  • The Lampo Group, LLC d/b/a Ramsey Solutions

Note: Just because a company got a letter, does not mean they (at this point) have done anything wrong.

However the letter also put those companies on notice.

Receipt of this notice of penalty offenses puts you and your company on notice that engaging in the conduct described therein could subject you and your company to civil penalties of up to $50,120 per violation.

We are aware of information suggesting that you have engaged in or are engaging in deceptive or unfair conduct. You should take prompt action, including by reviewing all your practices, to ensure any deceptive or unlawful claims cease and are removed or corrected, as appropriate, and any other required disclosures are made.

https://www.ftc.gov/system/files/ftc_gov/pdf/NPO-Misuse-Information-Collected-Confidential-Contexts-Cover-Letter_0.pdf

At the same time, the FTC published a blog post about the what it expects. It states that at minimum, companies which get confidential data need to get a consumers affirmative express consent prior to using that data for any purpose other what the consumer explicitly requested. The following paragraph has some examples worth noting:

[T]he Commission considers it an unfair or deceptive act or practice to use tracking technologies such as pixels, cookies, APIs, or SDKs to amass, analyze, infer, and transfer information collected in a Confidential Context for the purposes described in the prior paragraph without first obtaining affirmative express consent. It is also an unfair or deceptive practice to misrepresent or omit material facts regarding the use or confidentiality of information collected in a Confidential Context through tracking technologies such as pixels, cookies, or SDKs.

https://www.ftc.gov/business-guidance/blog/2023/09/companies-warned-about-consequences-loose-use-consumers-confidential-data

The blog is also quick to call out that:

Spoiler alert: burying something in your Privacy Policy or Terms of Service doesn’t meet the “clear and conspicuous” standard.

https://www.ftc.gov/business-guidance/blog/2023/09/companies-warned-about-consequences-loose-use-consumers-confidential-data

So it could be reasoned that the FTC is likely expecting some sort of overt banner, widget, dialog, or popup to be shown to the user prior to the collection taking place. It may also be a good idea to have a consent system of record – so it can be shown that a specific user actually did consent to the data usage.

Next Steps

Any company dealing with confidential contexts will want to take note of this warning. They’ll need to evaluate their data collection, determine if consent is needed and either build in systems to request such consent, or remove the offending technology. By the look of the letter, the FTC expects quick action, and there will be no further warnings.

The post FTC Warns Over Improper Data Collection appeared first on Development & Analytics.

  •  

Computer Security Requirements Surge

While I often talk about data privacy on this blog, something that has become a bigger focus in the content I read for the past few months has been the inclusion of increasingly rigorous security requirements when it comes to collection and processing of personal data. This is notable because much like data protection law is applying to entire new industries as these laws enter enforcement – these requirements are often paired with the need to hire a security service, or hire computer security specialists – sometimes both. Let’s take a look at how this is playing out with a glance at a few of the laws and proposed regulations.

Safeguard Rule

This past June the FTC began enforcement of the updated Safeguard Rule. These additional requirements require organizations subject to the rule to:

  • designate a qualified person to oversee their information security program,
  • develop a written risk assessment,
  • limit and monitor who can access sensitive customer information,
  • encrypt all sensitive information,
  • train security personnel,
  • develop an incident response plan,
  • periodically assess the security practices of service providers, and
  • implement multi-factor authentication or another method with equivalent protection for anyone accessing customer information.

Not only that, but the Rule was expanded to include more industries – so now the above requirements apply to industries typically exempt, such as Car Dealerships. In their own words:

Rather, the FTC Safeguards Rule covers businesses like mortgage lenders, mortgage brokers, motor vehicle dealers, payday lenders, finance companies, account servicers, check cashing companies, wire transferors, collection agencies, credit counselors and other financial advisors, tax preparation firms, non-federally insured credit unions, and investment advisors that aren’t required to register with the SEC. That’s not an exhaustive list, so if you aren’t sure if you’re covered, now’s the time to nail that down.

https://www.ftc.gov/business-guidance/blog/2022/11/compliance-deadline-certain-revised-ftc-safeguards-rule-provisions-extended-june-2023

This was paired with a notice from the Department of Education, which spelled out new enforcement requirements when it came to Student Information regarding financial aid. The key takeaway here – is that having a working Security program is now a requirement for many industries that had no such requirements prior to June, 2023.

SEC requires Cybersecurity Disclosures

While the Safeguard Rule didn’t apply to companies required to register with the SEC, the Securities and Exchange Commission had their own requirements which were finalized and will become required with annual reporting for fiscal years ending on or after December 15, 2023.

Under these new rules companies will be required to disclose any material cyber security incidents in 4 days (in most cases) on a Form 8-K and to describe their processes, if any, for processing, identifying and managing material risks from cybersecurity threats in Form 10-K.

These requirements will bring increased transparency regarding cybersecurity risks to publicly traded companies. It’s not hard to imagine that this may result in legal action, stock declines or bad press for companies found to be lacking in the cybersecurity realm.

California Enters the Ring

While the final regulations for the California Consumer Privacy Act were approved earlier this year, the Cybersecurity requirements were not among them. Over the past week we’ve seen the direction the California Privacy Protection Agency wishes to go with draft regulations (which may or may not become final). These nearly six pages of requirements are extensive, and would require companies subject to the CCPA to build and maintain a cyber security program and conduct audits by an external auditor, at least annually. These audit results(pass or fail) must be signed by a member of the board, or highest ranking executive with authority to bind the business.

A full review of the Cybersecurity draft can be found here.

Demand for Skills

All of these laws (and more not mentioned here) will need staff to implement and run their computer security program. According to NIST, there is a 3.4 million person global shortage of cyber security professionals. It seems likely that for the foreseeable future that demand for cyber security professionals will excessively outpace ability to fill roles. With all these new industries facing building out their programs at the same time, many may face delays due to lack of skilled staff.

All of this should be considered when grappling how to deal with the new reality of the pairing of robust security programs and data privacy programs which we will find ourselves in as these requirements spread across industries and geographical boundaries.

The post Computer Security Requirements Surge appeared first on Development & Analytics.

  •  

State Privacy Enforcement Steps Up

On July 1st, the States of California, Colorado and Connecticut had their respective data privacy laws become eligible for enforcement. In the weeks since we’ve already seen two of these States send out letters requesting information and warning that enforcement is beginning.

California

This year on January 1st, 2023, the employee data exemption for the California Privacy Protection Act (CCPA) expired. In light of this – the California Attorney General has sent letters to a number of letters to large California employers requesting information on how the companies comply with the law regarding handling of employee and job applicant data. It is very possible that enforcement action will result from this sweep – as California no longer has a cure period.

It should be noted that the California AG actively does enforcement sweeps of the CCPA. This was how Sephora was caught, and how mobile application developers came under investigation earlier this year.

Colorado

Meanwhile over in the Centennial State, employers in Colorado started to receive letters from the Colorado Attorney General educating companies on their new legal obligations. The AG was quick to warn that should they become aware of organizations (Colorado’s Act also applies to non-profits) they would be prepared to act for enforcement.

The letters will focus on obligations around collection and use of sensitive data, particularity around consent. Additionally in focus is the ability to allow consumers to opt-out of targeted advertising and profiling. Currently Colorado does have a cure period prior to enforcement, but this will sunset in the future.

Next Steps

Both of these States have issued press releases and effectively warned that enforcement is on the table going forward. Companies subject to either law should be aware that enforcement sweeps are a possibility and be prepared to respond quickly to any request for information. In the short term, ensuring your documentation is in order is likely a prudent course of action as timelines to respond may be shorter than expected, and effort to generate the documentation may be far larger then anticipated should it not already exist.

The post State Privacy Enforcement Steps Up appeared first on Development & Analytics.

  •  

Federal Agencies warn Healthcare Providers over HIPAA concerns

On July 20th, The Federal Trade Commission (FTC) and the Department of Health and Human Services (HHS) sent 130 letters to healthcare companies reminding them of their obligations around tracking technology and the Health Insurance Portability and Accountable Act (HIPAA). The letters indicate that the agencies are aware of possible privacy concerns in the recipients website or app, and encourages them to review and correct any possible issues. The letter specifically calls out that use of common marketing technology such as Google Analytics and the Meta pixel are very likely to be in violation of HIPAAs rules around impressible disclosure. The letters reflect a more targeted warning than the December Bulletin, and likely represents the final warning prior to enforcement action.

In recent times the FTC has been very proactive on enforcement over HIPAA violations, reaching settlements with companies such as  GoodRx, Flo Health, BetterHelp, and 1Health, In each case in addition to a fine and corrective action, the companies were often bound to a consent decree mandating specific behavior (such as not selling / sharing user data) going forward. This is to say nothing of the reputational harm and possible lost business due to break in consumer trust.

Those in the healthcare space are strongly encouraged to spend the time reviewing the letters, and considering how they may apply to their own efforts and taking corrective action as needed. Very likely – time is running out to get your house in order and the risk of enforcement is escalating.

The post Federal Agencies warn Healthcare Providers over HIPAA concerns appeared first on Development & Analytics.

  •  

Safari’s Advanced Privacy Protection

Back in June I briefly covered some of the changes to Safari with the upcoming operating system updated Apple had planned for the fall of 2023. Now that these updates have come, I wanted to get a deeper look at some of the files and behavior of Safari’s Advanced Privacy Protection features. Just how does Advanced Privacy Protection defend users?

How often is Private Mode Used?

This is a metric that is hard to measure, but Jason Packer over at Quantable gave it a try. Based on his testing the amount varies by both industry, operating system and form factor, which I found rather interesting but generally falls between 5 and 10%. It’s worth a read to fully understand the methodology.

Now what is important to realize is that while Advanced Privacy Protection is enabled by default in Safari’s Private mode – users who enter settings can enable it for all browsing. Since this requires user action of modifying default behavior – I would guess that for most, this would only increase the possible impact by a percentage point or two but this is important as you can’t assume that because a user isn’t in Private Mode everything will work.

It’d be worth understanding the possible impact to your specific data set before panicking.

Link Tracking Protection

The behavior of Link Tracking Protection hasn’t changed from my summer review, and the list of name / value pairs removed from the URL remains consistent with my list of vendors from July. A raw GIT file of the parameters can be seen here.

The affected vendors and their clients should begin to notice the impact from these changes in their reporting as of mid-September, when the iOS OS upgrade was released.

Tracker Domains

Apple identifies 627 domains as trackers in the TRACKING_DOMAINS.wplist file. This is the file that Safari appears to be using to determine if a network request should be routed through it’s Private Relay feature.

There’s also a TRACKING_SUBNET.wplist file – which has entries for Adobe, Criteo and Google.

Tracker Blocking

The tracker blocking is where it gets interesting.

There is a file known as URL_FILTER.wplist which has over 3800 entries. Each entry looks at a specific request type, does a RegEx match, determines if the request is third-party, and if so, performs an action – typically blocking the request outright. Some of the Regex appears to apply to specific companies usage of these systems where other strings are more far reaching to all companies which use the platform.

If you ever wondered who’s using specific vendors you may find that information in this file.

There’s some platform specific impacts that are worth mentioning.

Tag Managers

The gang’s all here. There are entries for Google Tag Manager, Adobe, Ensighten, Segment and Tealium.

In general – under Advanced Privacy Protection you can expect your tag managers to be blocked from loading. By extension anything they load will also be blocked from loading. If you are loading items like Consent Managers via your Tag Manager, it may be worth re-evaluating that decision.

Analytics Platforms

43 Entries return when searching for Regex targeting Analytics platforms, and this includes items such as Adobe Analytics, Google Analytics, Mix Panel, Amplitude, Heap and Site Improve.

Depending on how your loading your analytics – you can expect Advanced Privacy Protection to prevent data collection by blocking the required scripts from loading. There is also regex to apply to various Beacons or noscript pixel solutions.

Optimization Platforms

But what about A/B testing? Platforms which handle this service are also impacted with entries existing for platforms such as: Visual Website Optimizer and Monetate,

It seems that currently most Optimization platforms are not yet impacted.

Other callouts

There’s a lot of entries in this file – but a few other big names jumped out at me as likely having issues such as HubSpot & Treasure Data.

Next Steps

It’s worth testing how your site performs when viewed with Advanced Privacy Protection. Ideally, your site still works as intended, but if your developers got “clever” or didn’t practice defensive coding to account for the possibility a service may be unavailable then you may have issues that need to be addressed.

In general you’d want to understand your specific risk of adverse customer experience, as well as determine how any services you leverage are impacted (blocked, restricted feature access, etc.). From there you’ll be able to determine if these services are critical, and if so – determine how you may need to adjust your site to work without them, or to adjust the delivery of those services to be more compatible with the restrictions of Advanced Privacy Protection.

Advanced Privacy Protection is in public availability now, and adoption will continue to grow for the next several months as Apple pushes these changes out to users devices.

One final note worth mentioning is – these files can be updated by Apple at any time. Renaming files/paths to get around the Regex is, at best, a limited time option and effort may be better spent learning how to work with the new restrictions, rather than around them given Apple’s track record of closing loopholes created by industry.

The post Safari’s Advanced Privacy Protection appeared first on Development & Analytics.

  •  

Google Analytics in Europe – A Year Later

Let’s take a look at where things stand in regard to the use of Google Analytics (and data transfer in general) between the United States and Europe.

Legal Proceedings

Early 2022 was awash with news about Austria finding Google Analytics Universal Analytics violates Chapter V (regarding International Data Transfers) of Europe’s General Data Protection Regulation (GDPR) due to the Schrems II decision.  As the year continued to roll on, Austria was joined by decisions in France, Italy, and Denmark.  Other Data Protection Agencies have not been silent on the matter. Guidance has been issued out of  Norway and the Netherlands cautioning against use and recommending looking at alternative vendors.

It’s important to realize that while Google and Google Analytics have been the target of these investigations, they are not the only ones affected.   The Schrems II decision has far reaching impacts on any company that transfers data that may be covered by GDPR to the United States.

In response to this, The White House in the United States called for the creation of a Trans-Atlantic Data Privacy Framework to be created, which once agreed upon with the EU, would restore the ability to transfer data to the United States without violation of European law.  A draft was sent to the EU for consideration in late 2022.

Since then, the European Commission has issued a draft decision approving the measure and asked the European Data Protection Board for their opinion, which they have recently issued, citing several concerns with the proposed agreement.   The formal decision continues to work its way through the EU government with no expected decision date yet issued. However it is unlikely to be resolved before the spring of 2023.

Lastly the United States has FISA Section 702, which is one of the key concerns related to surveillance raised by the Schrems II decision, and it is up for renewal before the end of 2023.  Rather than let it expire, which I suspect would help the privacy agreement discussions, the Biden administration has opted to make a case for renewing FISA (Foreign Intelligence Surveillance Act) surveillance powers. I have to wonder if this will ultimately come back to hamper efforts the United States are making in seeking to restore international data transfers.

Google’s Response

Google, for its part, is still rather stuck in the middle of a legal dispute between the United States and the European Union with limited avenues for how they can resolve matters. Two events have started in the past year that move Google into a better (but not ideal) state.

First, they are sun setting Google Analytics Universal Analytics (which is the software mentioned in all of the DPA decisions) in favor of Google Analytics 4, which has more privacy protections built in.  However this is a long phase out process with it not being fully complete until July 1st, 2024.  While this helps position them long term, this date is still 15 months out at the time of writing, and that’s a long time for other Data Protection Authorities to reach similar decisions as they work through NYOB’s complaints.

Secondly, in the spring of 2022, they did upgrade their Google Analytics 4 product with a number of EU based privacy controls. Those controls likely do not go far enough to address the concerns raised by the Data Protection Authorities as cited by guidance from Denmark and Norway.   So while the decisions from the various countries revolved around website operator use of Universal Analytics, the changes made in support of Google Analytics 4 are likely to be found insufficient in altering a decision outcome in the future.

Conclusion

Things do not look positive in my opinion.   The agreement between the United States and EU over data privacy will likely continue to run into roadblocks, and even if ultimately successful, I suspect will be immediately challenged in the European Courts.  For its part, the United States is likely going to make things harder on themselves by seeking to renew Section 702 powers, which were part of the reason the international data transfer issue exists to begin with.

Google will continue to be stuck in a State level legal dispute which they are being impacted by, but can’t directly address.   The various complaints issued by NYOB will continue to be resolved, and Google and Facebook (which were the targets of the complaints) will likely continue to rack up adverse decisions regarding the use of their products.

Finally, Google Analytics 4, while an advancement in many ways, does not go far enough to address the privacy concerns and so I suspect the complaint decisions will find similarly in future cases.  This is unfortunate because Google, like other affected businesses, can’t fix the problems raised and so risk losing business from international clients in the meantime. 

The post Google Analytics in Europe – A Year Later appeared first on Development & Analytics.

  •  
❌