The proliferation of AI-powered offerings in the marketplace and discussions about AI's effect on jobs has become impossible to ignore. According to Goldman Sachs, 36% of S&P 500 companies mentioned AI on their earnings call in the last quarter of 2023.1 As proclamations of AI-powered capabilities infiltrate our personal and professional lives, the regulators have taken notice and started questioning the validity of these claims, with the Securities and Exchange Commission (SEC) leading the way.
What Is “AI Washing”?
- AI Washing is the new form of an old trick. Corporations that make deceptive claims to overstate their commitment to environmentally friendly practices are known to engage in greenwashing. Those who overstate their commitment to responsible social and economic practices without making any real policy reforms, on the other hand, are known to be bluewashing. As you may have guessed, AI Washing occurs when a corporation makes misleading claims about its AI capabilities.
AI Washing can take many forms. It can be as simple as exaggerating existing AI capabilities or go as far as simply lying about when and how a company uses AI. Making unfounded claims about the limitations and risks of how a corporation uses AI can also be considered AI Washing. During a panel discussion in December 2023, SEC Chairperson Gary Gensler warned corporations not to misrepresent their AI capabilities through AI Washing and stated that AI-related disclosures must be “full, fair, and truthful.”
Enforcement Actions
In a March 2024 speech at the American Bar Association (ABA) White Collar Crime Conference, Director of Enforcement Gurbir Grewal drew attention to emerging AI risks, from systematic financial risk to deception, hallucinations, and conflicts. Among those risks, notably, was AI Washing.
Not two weeks after Grewal’s comments, the SEC announced two settlements with investment advisors who made false and misleading statements about their use of AI. Below are summaries of each:
- Delphia (U.S.A.) Inc. – Delphia encouraged its clients to connect social media, banking, and other accounts to its trading platform, under the pretense that it would use AI and machine learning (ML) to incorporate the clients’ data to make robust investment decisions. Delphia claimed to be “the first investment adviser to convert personal data into a renewable source of investable capital.” The SEC found that Delphia had no such algorithms and never used clients’ data in this manner.
- Global Predictions, Inc. – Among several violations, Global Predictions advertised that its offerings included “expert AI-driven forecasts” and made bold claims regarding their models’ performance relative to benchmarks and historical returns. The SEC found that Global Predictions did not utilize “expert AI-driven forecasts” and that its claims regarding returns were misleading and unsubstantiated.
Key Takeaways
These two enforcement actions are likely the first of many and should serve as a warning to builders, maintainers, and investors of self-proclaimed AI-powered models across industries.
Here is what we will be watching for in the coming months:
- Additional focus on AI Washing and related AI concerns by the SEC resulting in additional settlements
- Other government agencies cracking down on AI Washing as well, including the Federal Trade Commission (FTC), the Consumer Financial Protection Bureau (CFPB), and various state-level regulators
- Companies will use more caution in the way they describe their use of AI/ML, and may perform pre-emptive reviews to ensure their marketing is consistent with their actual use of AI/ML.
- Based on the limited information included in the settlements, the Delphia and Global Predictions violations appear relatively straightforward: the companies claimed to use AI/ML but did not actually use it. The ultimate question will become more complex when companies are actually using some AI/ML.
- For example, what are the differences in how companies describe their use of AL/ML and the benefits to consumers and the actual construction of their AI/ML models and the measurable benefits?
- With the widespread use of AI applications, we are likely to see class action lawsuits based on AI Washing claims.
1 https://www.businessinsider.com/generative-ai-exaggeration-openai-nvidia-microsoft-chatgpt-jobs-investors-markets-2024-3
