Companies carry more liability for AI than they realize
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Sarah Grillo/Axios
Companies that make use of generative AI — whether they know it or not — could be taking on more legal liability than they realize, according to an article shared first with Axios.
Why it matters: The legal discussion around genAI has largely focused on the liability of the companies developing the technology rather than those who use it.
Driving the news: The 100-page article, published in NYU's Journal of Legislation and Public Policy, provides a detailed overview of the legal landscape in both the U.S. and around the world in terms of existing and emerging law.
EqualAI CEO Miriam Vogel, who co-authored the article, told Axios that many companies are under the false impression either that there is no liability for their use of genAI, or that the liability rests solely with the companies that build the technology.
- "That's just dangerous, because there are so many laws on the books that people need to know are applicable," Vogel said.
- The article was written by Vogel, former Homeland Security secretary Michael Chertoff, EqualAI legal director Jim Wiley and NYU law student Rebecca Kahn, a former EqualAI staffer.
Zoom in: While the U.S. has a lack of legislation specifically governing genAI, existing laws impose liability for outcomes when it comes to key issues.
- Regulators have made clear that existing laws around housing, lending and employment apply whether genAI was used or not.
- And even companies that think they aren't using AI because they haven't purchased or specifically authorized it may find that its unofficial use is widespread, sometimes with software not intended for commercial use.
- "And that's some of the AI that's least sophisticated, most rife with discrimination," Vogel said.
The big picture: The goal of the article isn't to dissuade people from using AI.
- "We want people to use it," Vogel said. "We want people to benefit. But it can only happen well if we go about it with our eyes open."
Between the lines: For many issues, the courts may have to determine how existing laws will be applied to genAI.
- That's especially true in the area of intellectual property, where artists, publishers and others are suing AI companies, claiming that both the training and operation of genAI models infringe their copyrights.
- Those who developed the models have argued they are protected by the notion of "fair use."
- On the flip side, it's unclear just how much human involvement is needed for works produced in conjunction with AI to be themselves eligible for copyright protection. The law has long held that information produced solely by computers is not eligible for protection.
