AI Risk Mitigation and Legal Strategies Series No. 4: AI Washing
This article was published by Legal Insight. https://www.legaldive.com/news/avoid-regulatory-risk-ai-washing-greenwashing-artificial-intelligence-FTC-SEC-scrutiny/704507/
Both SEC and FTC recently expressed their strong stand against AI washing. Is your company prepared to avoid becoming an example and facing substantial penalties? In this article, I have provided comprehensive and practical steps for in-house legal counsel to follow to ensure compliance with SEC and FTC rules.
a. Definition and Examples of AI Washing
"AI washing" is a deceptive marketing practice misrepresenting the extent of artificial intelligence (AI) capabilities in products or services.
The following are common AI washing examples:
1. Some companies may loosely use the term "AI" to describe basic automation or rule-based systems lacking authentic learning and adaptive capabilities.
2. Others may exaggerate the AI predictive capacities when the product primarily relies on simple statistical analyses or predetermined rules.
b. SEC Stance
On December 5, SEC Chair Gary Gensler cautioned public companies against AI washing. Gensler commented that public statements made by reporting companies about their AI capabilities fall squarely within the purview of securities regulations, which must be "full, fair, and truthful."
Gensler even compared AI washing to greenwashing, where companies made unfounded representations about environmental sustainability. The SEC has significantly ramped up enforcement actions against greenwashing. It recently fined Deutsche Bank's investment arm, DWS, $19 million for "materially misleading statements" relating to greenwashing in ESG funds.
In light of Gensler's remark and the SEC's enforcement efforts in greenwashing, it would not be a surprise that public statements of publicly traded companies promoting their AI products or services will be subject to stricter scrutiny.
c. FTC Stance
Section 5 of the Federal Trade Commission Act (FTC Act), which expressly prohibits unfair and deceptive trade practices, gives the FTC the authority to take enforcement action against AI washing. 15 U.S.C. § 45. The FTC attorney, Michael Atleson, provided the following guidance in his blog:
1. Be truthful when labeling your product as AI-powered. The FTC will investigate baseless claims and conduct a thorough analysis to verify the product's use of AI. It's important to note that using an AI tool in development does not automatically qualify a product as AI-powered.
2. Make truthful claims about your AI product and avoid exaggeration or assertions beyond current AI capabilities. Performance claims should have scientific support and be universally applicable, avoiding deception based on specific user types or conditions.
3. Exercise caution when claiming the superiority of your AI product over non-AI alternatives. The FTC requires substantial proof for comparative claims often used to justify higher prices or influence decisions. Only make such claims if obtaining adequate proof is possible.
4. Before launching your AI product, being aware of potential risks and their impact is crucial. In case of failure or biased results, attributing blame to third-party developers or claiming a lack of understanding or testing is unacceptable.
d. Legal Strategies
To comply with SEC and FTC rules, it would be prudent for an in-house legal counsel to consider taking the following steps:
1. Work closely with your tech team to analyze the AI technology utilized in your product or service. You may want to ask your tech team the following questions:
i. Does the technology itself have AI capabilities? If the answer is "No," and they claimed AI simply because they used AI tools to develop the product or technology, then the product is not AI-powered. If the answer is "Yes," ask the questions in items (ii) and (iii) below.
ii. Does a product primarily rely on simple statistical analyses or predetermined rules without incorporating more advanced machine learning techniques? If the answer is "Yes," then the product is not AI-powered.
iii. Is the technology a basic automation or rule-based system that lacks authentic learning and adaptive capabilities? If the answer is "Yes," then the product is not AI-powered.
If the answer is "No" to both items (ii) and (iii) above, then ask the questions in item (iv)- (vi) below.
iv. How does the AI system learn? Is it supervised learning, unsupervised learning, or reinforcement learning? What type of data does it learn from? What types of adaptations can the AI system make? Can it adjust its behavior based on new data or feedback? Can it improve its performance over time? Does scientific proof back up your performance claim?
v. How does the AI system handle unexpected or out-of-distribution data? Can it adapt to new situations, or does it require retraining?
vi. What are the potential risks if AI does not function as predicted? What safeguards are in place to prevent the AI system from making biased or discriminatory decisions? (I discussed the AI discriminatory issue in the Financial Services industry in this article: https://www.lklawfirm.net/blog/financial-services-aml-glba-fcra-ecoa-regulatory-compliance-artificial-intelligence )
2. Based on the answers to the questions above, you can determine if your product is genuinely AI-powered as well as AI's capabilities, limitations, and potential risks.
3. Work with the marketing department to review and approve marketing materials based on the first step's results and ensure all claims are supported by scientific proof.
4. Advise the executive team regarding the potential business and legal risks of AI technology used in the products or services. Consider distributing the costs and risks through an AI licensing agreement with clients.
5. Review and, if necessary, renegotiate the liability and indemnification provisions in the contracts with the vendors whose services impact your AI technology.
Please feel free to reach out to me at lkempe@lklawfirm.net if you have any questions or need assistance.
REFERENCES
Linthicum, David. From ‘cloud washing’ to ‘AI washing’. (March 28, 2023) https://www.infoworld.com/article/3691896/from-cloud-washing-to-ai-washing.html
Vanderford, Richard. SEC Head Warns Against ‘AI Washing,’ the High-Tech Version of ‘Greenwashing’. (December 5, 2023). https://www.wsj.com/articles/sec-head-warns-against-ai-washing-the-high-tech-version-of-greenwashing-6ff60da9
Karsh, Patrick. Don't AI Wash Investment Pitches. Medium. (March 15, 2023). https://patrickkarsh.medium.com/dont-ai-wash-investment-pitches-1fd9b54920e8
SEC. Deutsche Bank Subsidiary DWS to Pay $25 Million for Anti-Money Laundering Violations and Misstatements Regarding ESG Investments. SEC News Release No. 2023-194 (Sept. 25, 2023). https://www.sec.gov/news/press-release/2023-194
Atleson, Michael (February 27, 2023). Keep your AI claims in check. https://www.ftc.gov/business-guidance/blog/2023/02/keep-your-ai-claims-check
