0
An image of the flag of Europe, which consists of twelve golden stars forming a circle on a blue field.

EDPB Opinion on AI Models and GDPR Principles: Key Takeaways

In December 2024, the European Data Protection Board (EDPB) issued an Opinion in response to a request from the Irish supervisory authority, focusing on the application of GDPR principles in the context of AI models. The Irish supervisory authority posed three specific questions:
  1. When and how can an AI model be considered “anonymous”?
  2. What is the appropriateness of legitimate interest as a legal basis for AI deployment and development?
  3. What are the consequences of unlawful processing of personal data on subsequent operations of the AI model?
Through its answers, the EDPB provided key guidance on how AI models interact with fundamental rights to privacy and data protection established in the GDPR.

Anonymous AI Models

According the EDPB, “[f]or a model to be anonymous, it should be very unlikely 1) to directly or indirectly identify individuals whose data was used to create the model, and 2) to extract such personal information from the model through queries.” While anonymous data can help mitigate privacy concerns, it does not automatically make the AI model completely exempt from GDPR compliance. When a model is claimed to be anonymous, supervisory authorities will evaluate the claims of anonymity on a case-by-case basis, considering “all the means likely to be used” by the controller or a user. The Opinion states that supervisory authorities should review the documentation provided by the controller when assessing if the model is truly anonymous. The EDPB outlines methods that the controller may use to demonstrate anonymity, which may include: 1) reducing the amount of personal data used during training, 2) taking steps to ensure this data cannot be identified, and 3) utilizing technical safeguards to prevent data extraction from the AI model using prompts or queries. Key Takeaway: If a business claims an AI model relies on anonymous data, the claims of anonymity should be substantiated on a case-by-case basis with sufficient evidence and documentation. To do this, businesses with allegedly anonymous AI models may need to implement technical measures to limit the collection of data, reduce the likelihood of data being identifiable, protect against that data being extracted by users during deployment, and create documentation capable of demonstrating these efforts.

Legitimate Interest as a Legal Basis

Under the GDPR, a legitimate interest may constitute a legal basis for companies to process personal data when they have a justifiable reason to do so (beyond obtaining consent). However, the legitimate interest should be balanced against the data subject’s rights and interests, which requires careful consideration and justification when processing information from data subjects. The Opinion provides a framework to assess if a legitimate interest can be a valid legal basis for processing personal data in AI development and deployment. The framework is comprised of a three-step test:
  1. Identify the legitimate interest pursued by the controller;
  2. Assess the necessity of the processing for purposes of the legitimate interest; and,
  3. Balance the legitimate interests against the rights and freedoms of the data subjects.
When conducting this test, the controller should be careful to identify an interest that is lawful, clearly articulated, and non-speculative. For example, a legitimate interest may be to develop an AI model’s conversational agent or to improve threat detection in an information system. The controller should also adhere to GDPR data minimization principles, which state that the processing activities must be proportionate and in line with only what is necessary to achieve the legitimate interest. Finally, controllers should conduct a nuanced balancing test. This test considers the unique circumstances of each case, which may include the data subject’s interest in retaining control over their data, personal benefits, or socioeconomic interest. The Opinion notes, the more precisely an interest is defined in relation to the purpose of the processing, the more precise the estimation of benefits and risks will be. By employing this framework, developers and deployers should be able to decrease the likelihood that their AI models are disproportionately infringing on individual privacy rights and better align their AI practices with GDPR requirements. Key Takeaway: The three-step analysis, according to the Opinion, is crucial to improving compliance for organizations relying on legitimate interest as a legal basis for processing in AI development or deployment. Organizations relying on legitimate interest in this AI context should review their processing activities to determine whether they are proportionate, transparent, and aligned with GDPR principles—like data minimization—to justify the reliance on legitimate interest as a legal basis for processing.

Consequences of Unlawful Processing

The Opinion notes that supervisory authorities enjoy discretionary powers to investigate and assess violations, and they can choose appropriate remedial measures based on the context of the case. However, the EDPB also provides guidance for the supervisory authorities, based on three scenarios.
  1. In the first scenario, personal data is retained in the AI model. The Opinion states that supervisory authorities will need to consider the surrounding circumstances of the AI model to determine if the development and deployment phases of the model involve different legitimate purposes for processing. If so, each should be examined separately.
  2. In the second scenario, personal data is retained in the model and is processed by another controller during deployment. In this instance, the supervisory authorities should determine if the deploying controller conducted an appropriate assessment to demonstrate accountability with Articles 5(1)(a) and 6 of the GDPR. This assessment should show that the AI model was not developed by unlawfully processing personal data.
  3. In the final scenario, a controller unlawfully processes personal data to develop the AI model, and then anonymizes the data before processing it in the context of deployment. The Opinion states that, if it can be demonstrated to the supervisory authorities that the deployment of the AI model does not entail the processing of personal data, then the GDPR does not apply. Therefore, the unlawfulness of the initial processing in development should not impact the deployment operation of the model.
While supervisory authorities do have substantial discretion in oversight of processing activities, the scenarios highlighted by the EDPB show that the development and deployment phases, while connected, may need to be evaluated independently. Key Takeaway: Organizations should proactively ensure compliance at both the development and deployment stages of an AI model. Supervisory authorities will likely use the above examples as guidance, emphasizing the important of demonstrating lawful practices through each stage of the model. The EDPB’s Opinion is an important guide for organizations navigating the intersection of AI and data privacy law. By addressing issues around anonymous AI models, legitimate interest, and lawful processing in development and deployment stages, the Opinion emphasizes responsible AI development. As AI technologies continue to advance, businesses should be aware of the ways supervisory authorities are overseeing their AI models. The insights provided by the EDPB provide a foundation to help businesses to advance and develop new AI models, while also helping to safeguard and protect the rights of individuals.
0

Metaverse Law in Orange County Lawyer Magazine

The January 2025 edition of Orange County Lawyer magazine features an article written by Metaverse Law’s Lily Li. Read “AI and Machine Learning in Drug Development and Clinical Trials” below or in Orange County Lawyer magazine.
[Originally published as a Feature Article: AI and Machine Learning in Drug Development and Clinical Trials, by Lily Li, in Orange County Lawyer Magazine, January 2025, Vol. 67 No.1, page 28.]   AI and Machine Learning in Drug Development and Clinical Trials by Lily Li   In 2013, sleep medication zolpidem (Ambien, Ambien CR, and Edluar) swept headlines. Marie Claire reported on an alarming and suspicious rise in users experiencing irrational eating, gambling, and even “sleep-driving” while in a hypnotic trance—waking with no memories of their actions.[1] In several cases, women arrested and convicted for driving under the influence contested their convictions, arguing that they were not liable for these undisclosed drug-related side effects. At the same time, several clinical studies suggested that women metabolized zolpidem differently from men. By reviewing existing literature, Japanese researchers out of Shimane University identified 40% higher concentrations of zolpidem in women than men following use, and higher rates of visual hallucinations and sensory distortions.[2] The FDA released a safety advisory, warning users of the risks of “next-morning impairment” for the use of Ambien and related drugs.[3] In addition, the FDA took the unusual step of recommending a 50% cut in the dosage for women. When asked about the change, an FDA director told ABCNews.com: “The changes are different in women and men . . .We don’t understand why yet, but women are more susceptible to next-morning impairment.”[4] Yet, a decade later, the evidence supporting different zolpidem dosages for women and men is unclear.[5] In part, this is due to the lack of research surrounding sex differences in drug impact and drug treatment, as well as substantial gaps in the inclusion of women in clinical studies. From 1977 to 1993, FDA policy recommended excluding women of childbearing potential from Phase 1 and early Phase II drug trials.[6] Even after this policy was removed in 1993, industry fears remained with respect to drug interactions with pregnancy. This episode with zolpidem raised several concerns in the drug development and clinical trial process:
  • How do we recruit representative candidates for drug trials?
  • How do we ensure the quality and availability of datasets for clinical research?
  • How do we measure potential impacts of drug dosing on different populations?
  • What are the legal implications for failing to address appropriate drug doses?
  AI and ML to the Rescue? Now that artificial intelligence is being used in research and development, one wonders: Can artificial intelligence (AI) and machine learning (ML) reduce bias and risks during drug development? Or will it create new legal risks due to bias, privacy intrusions, and lack of transparency? The FDA released a discussion paper on AI, Using Artificial Intelligence and Machine Learning in the Development of Drug and Biological Products, to discuss potential regulatory frameworks to address the use of AI and ML.[7] In this discussion paper, the FDA released a set of fascinating case studies into existing research and uses of AI in the clinical trial process. Several of these case studies are discussed below, as well as an analysis of their potential impact on the zolpidem example.
  1. Recruitment. According to the FDA, “AI/ML is being used to mine vast amounts of data, such as data from clinical trial databases, trial announcements, social media, medical literature, registries, and structured and unstructured data in EHRs [electronic health records], which can be used to match individuals to trials (Harrer, 219 Shah, Antony, & Hu, 2019).” In this manner, researchers can combine huge quantities of publicly available data and individual health data from prior research to identify participants with certain medical conditions (or lack of adverse conditions) for investigational treatments. For zolpidem, the use of AI/ML may have been able to identify a much broader list of participants for initial clinical testing, making it easier to assess and identify adverse reactions.
  2. Selection and Stratification of Trial Participants. In addition to initial recruitment, AI/ ML has the capability improve intake, selection, and classification of clinical trial participants. Based on baseline characteristics selected by the researchers, such as prior clinical data, and vitals/labs taken during intake, predictive algorithms can help identify high-risk participants.[8] These groups can then be randomized and then subject to more strict monitoring protocols. In the case of zolpidem, alcohol use is associated with sometimes severe adverse effects from the drug, and so it would be beneficial to screen out candidates with a history of alcoholism or, on the flip side, assess drug interactions for this high-risk group with additional support, monitoring, or counseling.
  3. Dose/Dosing Regimen Optimization. AI/ML can be used to predict drug exposure for different populations based on factors such as weight, height, sex, and other characteristics that might impact drug metabolism. Based on prior drug exposure and response profiles for similar drugs and similar populations, AI/ML can help to narrow the dose/dosing regimen selected for a study. As noted by the FDA’s discussion paper, this can help optimize drug dosing “in special populations where there may be limited data (e.g., rare disease studies, pediatric and pregnant populations).” Based on this research, we can imagine future scenarios where AI/ML could have avoided zolpidem dosing concerns, where graduated and limited dosing was tested and applied to different sex, age, and metabolism categories to determine ideal dosing.
  4. Data Analysis. On a more intriguing level, the FDA AI discussion paper discussed the concept of creating “digital twins” of patients for clinical trials. Essentially, an AI version of the clinical participant is created, using the existing candidate’s electronic health records, vital signs, labs and other records. Researchers can assess how the digital twin would react under normal conditions using AI/ML modeling based on data gathered from similar individuals. This digital twin would then act as a substitute for a placebo candidate in a clinical trial, and act as a benchmark against the actual patient undergoing investigational treatment. For zolpidem, this could be used to assess candidates that already have underlying medical conditions such as anxiety, depression, or other confounding factors, to see whether an adverse effect from a trial is due to the investigational treatment or something that is likely to occur to the same individual from anxiety alone.
  5. Postmarketing Safety Surveillance. Finally, AI/ML can help detect and assess adverse events once the drug enters the market. This is not just limited to individual case safety reports (ICSR), required by regulators, but can include adverse events reported publicly on social media and the wider internet. This type of postmarketing safety surveillance could assist researchers and drug companies in identifying potential drug risks, prior to landing on primetime news.
  Quality and Reliability Risks While AI/ML can help to address the costs and efficiency of clinical trials, this relies substantially on the underlying data used to train AI. The quality and reliability of any AI/ML model requires similar quality controls for underlying training data. Given the safety risks of inappropriate drug dosing, or recruiting candidates with severe medical conditions, AI developers cannot rely solely on self-reported healthcare data with no external medical testing or validation. Developers should be equally wary of training on third-party data sets that do not provide documentation on the collection of data and data validation. Within an existing healthcare organization, if the organization is big enough, aggregate and de-identified data may be obtained from existing electronic health care records and prior clinical trials. Yet, even within these large datasets, errors may surface during training. Medical providers may code the same procedure, and similar symptoms, a dozen different ways. Even drug names can be misspelled and coded incorrectly within existing records. While many of these errors may end up being statistically insignificant with enough data, there is the risk of missing one or two major adverse events, or “black swan” events, that would otherwise change the entire risk profile of a drug. In addition to quality and reliability, the underlying dataset needs to be representative of the population that will be studied for the clinical trial. If the underlying dataset is only trained on a handful of individuals with a certain medical predisposition, age, sex, weight, etc., it will be difficult for the AI model to make predictions for that group. As an example, if the training data only contains the medical information for two individuals over the age of sixty, and shows no adverse effects from a particular drug dose, this information is not enough to generalize that the drug at that dosage is appropriate for all individuals over the age of sixty. For all we know, these two candidates could be a former Olympic diver and a nutrition coach, two outliers that completely skew the data. Consequently, the underlying training data for any AI model should also be assessed for bias and representativeness as it applies to the proposed clinical trial.   Data Privacy, Cybersecurity, and AI Risks The data privacy and cybersecurity risks associated with the foregoing uses of AI/ML cannot be underestimated. The quality and representativeness of any AI system in this field will rely heavily on large swathes of healthcare data, fine-tuned and, at times, personalized in the case of digital twins. This is sensitive or special category data at its finest, triggering heightened scrutiny under the EU’s data privacy law, the GDPR, and U.S. data privacy and data breach laws. To date, most healthcare organizations have sidestepped data privacy concerns by relying on HIPAA’s de-identification standard to remove personal information and other identifiers from healthcare data, making it difficult to associate with an individual. While the FDA requires Institutional Review Board (IRB) review of most biomedical research involving human subjects, this generally does not apply to de-identified personal information that cannot be linked to an individual. Simply de-identifying data and then running with it is not enough, however. Under the California Consumer Privacy Act and similar state laws, for example, recipients of de-identified data need to affirm that they will not attempt to reidentify the data (except to test their de-identification methods). The GDPR has a much higher “anonymization” standard, which looks at the re-identifiability of personal information, given all the different datasets that an organization may have access to. AI/ML itself is making the de-identification process harder. As it is capable of slicing and dicing data by age, race, sex, and medical condition, and combining multiple large datasets, it is easy to run the risk of re-identifying data. While several thousand people might have the same configuration of eye color, age, gender, and weight, only one or two may have participated in a clinical trial at a particular location, or have specific allergies or side effects to certain types of medication. As a result, in circumstances where healthcare data is not de-identified, or the risk of reidentification is heightened, then it behooves clinical organizations and their AI developers to implement written information security programs and associated privacy and security controls.   Legal Liability and Drug Dosing In several notable cases, defendants on zolpidem were able to contest or overturn DWI or even vehicular manslaughter cases. Essentially, these defendants argued that they were not aware of the potential dangers of zolpidem, and so could not be liable for their actions while “sleep driving.” This raises the question: If AI gets good enough, and can tell you exactly the right dose to take of a drug, will you (or your doctor) be liable if you deviate from the AI’s recommendations? Will the AI’s recommendations be discoverable in court (and surfaced via AI-enhanced search)? Only time will tell what this brave new world will bring.   ENDNOTES [1] Kai Falkenberg, While You Were Sleeping (September 27, 2012), Marie Claire, https://www.marieclaire.com/culture/news/a7302/while-you-were-sleeping/.   [2] Takuji Inagaki, Tsuyoshi Miyaoka, Seiichi Tsuji, Yasushi Inami, Akira Nishida, and Jun Horiguchi, Adverse Reactions to Zolpidem: Case Reports and a Review of the Literature, 12 Prim Care Companion J Clin Psychiatry 6 (2010), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3067983/.   [3] U.S. FDA, Drug Safety Communication: FDA approves new label changes and dosing for zolpidem products and a recommendation to avoid driving the day after using Ambien CR (May 14, 2013), https://www.fda.gov/drugs/drug-safety-and-availability/fda-drug-safety-communication-fda-approves-new-label-changes-and-dosing-zolpidem-products-and.   [4] FDA: Cut Ambien Dosage for Women, ABC News (January 10, 2013, 6:03AM), https://abcnews.go.com/Health/fda-recommends-slashing-sleeping-pill-dosage-half-women/story?id=18182165.   [5] David J Greenblatt, Jerold S Harmatz, & Thomas Roth, Zolpidem and Gender: Are Women Really At Risk?, 39(3) J. Clinical Psychopharmacol. 189 (May/Jun 2019), https://pubmed.ncbi.nlm.nih.gov/30939589/.   [6] NIH Inclusion Outreach Toolkit: How to Engage, Recruit, and Retain Women in Clinical Research, last accessed September 16, 2024: https://orwh.od.nih.gov/toolkit/recruitment/history.   [7] FDA, Using Artificial Intelligence and Machine Learning in the Development of Drug and Biological Products (May 10, 2023), https://www.fda.gov/media/167973/download; see also Using Artificial Intelligence and Machine Learning in the Development of Drug and Biological Products; Availability, 88 FR 30313 (May 11, 2023), https://www.federalregister.gov/documents/2023/05/11/2023-09985/using-artificial-intelligence-and-machine-learning-in-the-development-of-drug-and-biological.   [8] Thi Tuyet Van Tran, Hilal Tayara, and Kil To Chong, Artificial Intelligence in Drug Metabolism and Excretion Prediction: Recent Advances, Challenges, and Future Perspectives, 15 Pharmaceutics. 1260 (Apr 17, 2023), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10143484/.   Lily Li is an AI, data privacy, and cybersecurity lawyer and founder of Metaverse Law. She is a certified information privacy professional for the United States and Europe and is a GIAC Certified Forensic Analyst for advanced incident response and computer forensics. She can be reached at info@metaverselaw.com.
0

Upcoming Webinar: The EU AI Act: How Will It Affect Your Business?

On Thursday, December 12, 2024 at 11am ET, DataCamp will host a live webinar titled “The EU AI Act: How Will It Affect Your Business?” The speakers are Dan Nechita, Lead Technical Negotiator for the EU AI Act on behalf of the European Parliament, and Lily Li, Data Privacy, Cybersecurity & AI Lawyer and Founder of Metaverse Law. They will cover AI governance and risk management requirements under the EU AI Act and new US AI law. Attendees will:
  • Learn about the scope and requirements of the EU AI Act and other AI legislation.
  • Understand the risk classification system and the requirements for AI literacy.
  • Learn how to comply with regulations and the consequences of non-compliance.
  Join us for this informational webinar by registering at the DataCamp website using this hyperlink.
0
Photo of a judges gavel and block next to each other.

CCPA Board Meeting: Key Takeaways from November 8, 2024

In a vote of 4-1, the California Privacy Protection Agency (CPPA) has decided to move forward with rulemaking of its draft regulations concerning AI, cyber audits, profiling and risk assessments, despite complaints of regulatory overreach.

 

On Friday, November 8, the CCPA held a public meeting to discuss proposed updates to the California Consumer Privacy Act (CCPA) regulations. The hybrid meeting included public comments from a broad range of stakeholders – nearly 45 public comments were heard from business representatives, privacy advocates, and industry experts. While the passing vote would have typically triggered a 45-day public comment period on the draft regulations, Chairperson Urban requested flexibility, considering the upcoming holidays.

 

Legal Challenges

During the meeting, the CPPA stated that it was sued for failing to promulgate regulations, specifically on opt-out rights of information processed by automated decisionmaking tools (ADMTs). At the same time, commentators argued that the breadth of the proposed rules overstepped the intent of the CCPA.

 

Board Member Alastair Mactaggart–who helped draft the CCPA–voiced concerns about the regulations, arguing that the current proposed regulation is excessively broad to the point of being unworkable. He pointed out that these regulations, as written, apply to nearly all businesses that use any kind of software to generate any type of output–whether it’s AI-powered or not. For example, a simple tool like a spreadsheet or a school admission application could fall under these rules, forcing a large swath of low-risk businesses to conduct risk assessments. Mactaggart referred to this as statutory overreach and claimed that regulations should be focused on issues that genuinely impact privacy or security.

 

Economic Forecasts

The CPPA also issued a Standardized Regulatory Impact Assessment (SRIA) which was discussed during the meeting. In this assessment, the CPPA estimates the total cost of this regulatory initiative to be around $3.5 billion for the first year of implementation, with an average of $1 billion each subsequent year for the first ten years. The CPPA justifies this cost, asserting that the direct benefits to California businesses will be $1.5 billion in 2027, and $66.3 billion in 2036.

 

However, the California Chamber of Commerce states that “[b]usinesses, consumers and governments in California will suffer net losses from the proposed rules pending before the [CPPA] this week.” This statement stems from a report prepared for the Chamber of Commerce by Capitol Matrix Consulting, which concludes that the regulations are likely to “result in a substantial net losses to businesses, consumers, and governments in this state, both in the near and long term.”

 

Industry groups including TechNet, the Civil Justice Association of California, and the Interactive Advertising Bureau voiced concern about the heavy compliance burden that regulations place on businesses–especially small businesses that may not have the recourses to implement the required risk assessments or redesign their services to accommodate opt-out provisions.

 

Behavioral Advertising & Opt-Out Provisions

Another key point of contention during the meeting was the opt-out provision for consumers related to decisions made by AI systems.

 

The draft regulations govern a large range of AI. Under the draft, AI is defined as a “machine-based system that infers, from the input it receives, how to generate outputs that can influence physical or virtual environments.” Additionally, the draft defines ADMTs as “any technology that processes personal information and uses computation to execute a decision, replace human decisionmaking, or substantially facilitate human decisionmaking.”

 

Together, these definitions are more expansive than the definition of the high-risk automated processing addressed in Article 22 of the EU’s GDPR, the source of the original opt-out language. Under Article 22, a consumer has the right to opt out of decisions made by solely automated systems. The intent of this provision is to give consumers the ability to opt out of decisions that may be made on solely automated processes, such as targeted advertising.

 

However, critics argue that including the opt-out language in the draft in combination with an expansive definition of AI and ADMTs could have unintended consequences, especially for small businesses. Mactaggart, for instance, is concerned that applying this opt-out rule too broadly could lead to a breakdown of essential services. For example, online booking services for airlines and automated reservation software for hotels may rely on software that would be categorized as “AI” under this definition. Allowing users to opt out of using AI when asking for these services may be untenable, which could cause friction in these industries and ultimately could cause harm to consumers by limiting access to these services or increasing costs.

 

Risk Assessments

A central component of the draft regulation is for businesses who use AI, as defined above, to conduct risk assessments. While the goal of this requirement is to ensure that businesses are aware of and mitigate any potential privacy risks that arise from these technologies, critics believe the regulations go too far by applying the requirement to low risk, everyday activities.

 

For example, a representative from the California Grocery Association expressed concerns about how the opt-out provision would impact a chain of small rural grocery stores with whom she conducts business. While these AI tools could be used to help consumers save money, the cost of compliance to integrate these tools might not be within reach, especially given the thin profit margins within the grocery industry.

 

Again, Mactaggart questioned the scope of the draft. He and other advocates called for a narrower focus for risk assessments that centers on significant decisions–such as those that deny individuals access to essential goods and services. This could include the denial of a loan application, exclusion from an online platform, or an adverse employment decision. One commenter stated that there have been no public comments against regulating high-risk systems, and by focusing on these issues, the CPPA could better mitigate potential harms. At the same time, this would free low-risk systems from potential overregulation.

 

Additionally, a commentor suggested that risk assessments should be streamlined and aligned with other state standards to reduce compliance costs.  Mactaggart notes that accepting risk standards from other US jurisdictions could help businesses avoid duplicative efforts, cut compliance costs, and reduce the overall regulatory burden.

 

AI Training

The ability to opt out of training for AI datasets was of lesser concern but was still addressed by a number of commentors. For example, a representative from the Software and Data Industry Association argued that requiring an opt-out from consumers from AI dataset training could create a substantial burden on small businesses who already have trouble accumulating representative training data. Other commentors shares concerns that these opt-outs could compromise the quality and effectiveness for AI systems.

 

Ultimately, California faces a delicate balance in regulating AI and ADMT. On one hand, the state must work toward protecting consumers from privacy risks, potential discrimination, and other adverse impacts of AI. At the same time, the CPPA must ensure that rulemaking does not stifle innovation, create excessive compliance costs, or diminish competition between businesses that rely on AI.

 

As formal rulemaking moves forward, it will be crucial for the CPPA to consider feedback from the public comment period and to refine the regulations to ensure that they strike a balance between privacy concerns and costs to consumers and businesses alike.

0
Robotic hand and human hand pointing toward each other with the letters "AI" in between them.

Comparing EU and US AI legislation: déjà vu to 2020

This article was initially published in Reuters and Thomson Reuters Westlaw Today.   Lily Li of Metaverse Law discusses the landscape for AI legislation, with the passage of the European Union’s AI Act while states pass AI bills with differing thresholds, coverage and subject matter.   The landscape for EU and US AI legislation feels like a rinse and repeat of data privacy legislation in 2020. Back then, the General Data Protection Regulation (GDPR) was in full force and effect, while California and other states were developing privacy laws at breakneck speed. Many companies were caught unaware by GDPR, only to face a new onslaught of US state-by-state privacy laws.   Now, companies face the same problem. The EU has just passed a comprehensive AI law, the EU AI Act, which imposes significant compliance obligations and antitrust-style mega fines.   In the United States, state legislatures are passing AI bills at a breakneck speed, with differing thresholds, coverage and subject matter. Do global companies bite the bullet and comply with the EU AI Act globally, or should there be a more nuanced jurisdiction-by-jurisdiction approach?   Comprehensive and imposing   The EU AI act is a comprehensive law that has been in development for years by EU regulators. One of its unique features, not seen in US legislation, is a complete ban on certain “prohibited AI practices” (Article 5, https://bit.ly/4gQHfe8). Some of these prohibited practices include assessing whether an individual is likely to commit a crime and real-time biometric identification by law enforcement (think Minority Report), as well as social scoring of individuals.   In addition to setting forth prohibited practices, the EU AI Act designates a list of high-risk AI practices. This includes, but is not limited to, use of AI in employment decisions, credit scores, insurance and access to services. For these high-risk AI practices, AI providers need to implement a full risk management program that considers the following factors:  
  • Data governance
  • Technical documentation
  • Recordkeeping
  • Human oversight
  • Accuracy, robustness, and cybersecurity management
  • Quality management
  Like the GDPR, the EU AI Act imposes significant fines. This can be up to $35,000,000 or 7% of total worldwide revenue, whichever is higher, for engaging in prohibited AI practices (Article 99, https://bit.ly/3XRewgl), and up to $15,000,000 Euros or 3% of the total worldwide annual turnover, whichever is higher for other violations (Article 99, https://bit.ly/3XRewgl). The law requires each EU country to designate at least one independent and impartial body to monitor and enforce the EU AI Act’s requirements.   In contrast, the US is following a patchwork approach. Instead of comprehensive federal legislation, we are seeing a state by state and agency approach. To date, these laws generally fall into four main categories: (i) consumer protection; (ii) employment rights; (iii) image and likeness rights; and (iv) transparency/ risk assessment requirements for high-risk AI processing.   Consumer protection   For state consumer protection laws governing AI, Utah is one of the first movers. In May of 2024, it added requirements governing AI to its consumer protection statutes. Utah’s AI Policy Act requires businesses in Utah to disclose the use of generative AI tools, and also makes businesses liable for any consumer protection violations by these generative AI tools.   At the federal level, the FTC has used its consumer protection authority under Section 5 of the FTC Act, in order to regulate against unfair and deceptive practices in commerce concerning AI. In 2022, Weight Watchers agreed to pay a $1.5 million civil penalty in a settlement with the FTC, in part over allegations that the company improperly collected children’s data to train its models and algorithms. This settlement included “algorithmic disgorgement” — i.e., Weight Watchers was required to delete any models trained on such data.   More recently, on Sept. 25, 2024, the Federal Trade Commission (FTC) has cracked down on companies that make misleading or fraudulent claims about their use of AI tools. This included taking action against DoNotPay (https://bit.ly/3BtSWXW), a company that claimed to offer an AI service that was “the world’s first robot lawyer.”   DoNotPay agreed to a $193,000 settlement with the FTC, pursuant to a consent order. The consent order (https://bit.ly/4dNyjmN) also requires DoNotPay to refrain from “representing that its Service or any other internet-enabled product or service that it offers operates like a human lawyer or any other type of professional, unless that representation is not misleading and DoNotPay possesses competent and reliable evidence to substantiate the representation.” In addition, DoNotPay is required to notify consumers of the order and to submit compliance reports to the FTC.   AI in employment decisionmaking   At the employment level, Illinois recently enacted a law that prohibits the use of AI systems from discriminating against employees or job applicants based on any protected classes.   In addition, this amendment explicitly bans the use of race or zip code when used as a proxy for race in AI systems making employment decisions. Illinois’ requirements join New York City Local Law 144 (https://on.nyc.gov/3zHlSva) in regulating automated employment decision-making tools. While Local Law 144 does not include an explicit ban on the use of race or zip code in AI systems, it has very stringent notice and audit rights.   Where employers use AI systems “to substantially assist or replace discretionary decision making,” Local Law 144 requires publicly available third-party bias audits of automated employment decision-making tools.   Image and likeness rights   Generative AI is also regulated by state laws and cases governing image and likeness rights. Following the actors and writers strike in Hollywood, and high-profile litigation by Sarah Silverman and others, California has acted. In the last week, Governor Gavin Newsom signed two AI bills designed to protect entertainers.   AB 2602 requires contracts with actors and other performers to specify whether generative AI will be used to create a replica of the performer’s voice or likeness. AB 2836 bans the use of digital replicas for deceased performers, without the consent of the performer’s estate.   Transparency and risk assessment   The majority of US state comprehensive data privacy laws require transparency concerning the use of AI to process personal data and make decisions that impact important rights, such as employment, housing, and access to services. In addition, these laws generally give consumers the right to opt out of such processing.   Colorado’s AI Act, slated to go in effect in 2026, goes even further. It imposes risk assessment and bias assessment requirements for any “high-risk artificial intelligence system” that makes or is a substantial factor in making a consequential decision.   For purposes of the law, “consequential decision” means a decision that has a material or similarly significant effect on the provision or denial to any consumer of, or the cost or terms of:  
  • Education
  • Employment
  • Financial or lending services
  • Essential government services
  • Health-care services
  • Housing
  • Insurance
  • Legal service
  The Colorado AI Act has even more substantial transparency and notification obligations. As just one example, developers and deployers of “high-risk” AI systems are required to publicly post on their websites a description of the high-risk systems, as well as describe how the AI system manages the risks of bias. This includes further reporting to the Attorney General of “any known or reasonably foreseeable risks of AI discrimination arising from the intended use of the system.” Section §6-1-1702(5).   Where to go from here?   The trend lines are clear, and AI legislation is here to stay. While the US has not enacted federal AI legislation of the same scope as the EU AI Act, we already see significant risk assessment and transparency requirements. As a result, AI companies need to go global with their AI risk management strategies and not get left behind.   Lily Li is the founder and president of Metaverse Law. She advises global clients on their AI risk assessments and data protection impacts assessments, and supports her clients’ overall governance, risk, and compliance (GRC) programs. In addition, she holds the GIAC Certified Forensic Analyst (GCFA) certification for advanced incident response and digital forensics and certifications in information privacy such as the FIP, CIPP/US/E/M. She is based in Newport Beach, California, and can be reached at info@metaverselaw.com.
1 … 3 4 5 6 7 … 9