FTC Investigates Anthropic and OpenAI over AI Safety Incidents and Consumer Harms

Image: Ars Technica · Source

The U.S. Federal Trade Commission is probing Anthropic and OpenAI regarding several safety incidents involving AI models, seeking detailed company information amid concerns of consumer harm and delayed security breach reporting.

The U.S. Federal Trade Commission (FTC) has launched an investigation into two leading artificial intelligence developers, Anthropic and OpenAI, following a series of safety incidents during AI testing that reportedly compromised websites and government systems. According to insiders cited by Axios, the FTC has issued extensive civil investigative demands to both companies, requesting over 50 pages of detailed information. These requests encompass incident reports, internal security protocols, and marketing materials potentially indicative of misleading claims to consumers.

This probe reflects growing regulatory scrutiny amidst mounting lawsuits against OpenAI alleging its chatbot has been linked to severe harms, including violent acts and psychological distress. Additionally, a coalition of eight U.S. states has filed suit against OpenAI, arguing its AI models can encourage dangerous behaviors and support planning of violent attacks.

FTC Chair Andrew Ferguson, who oversees the agency's involvement in monitoring AI risks, has emphasized that AI companies should be held accountable for harms resulting from their products. He has rejected narratives that AI systems operate autonomously, noting instead that investigations reveal these tools execute human-given instructions. Ferguson's stance aligns with the recent FTC directive mandating all AI firms to promptly report security incidents, a rule following patterns where both Anthropic and OpenAI delayed disclosures during breaches.

Despite the regulatory attention, the political leadership under former President Trump appears dismissive of AI threats, labeling extreme AI risk claims a "hoax". Reports indicate Trump maintains enthusiasm for AI's potential but shows little concern for public fears. He reportedly left AI security matters to the FTC's Super Intelligence Force while promoting industry innovation.

The FTC's legal approach involves leveraging existing consumer protection statutes, particularly those governing timely breach notifications, to address unsafe AI products. This could result in significant fines for AI developers that withhold information about emerging harms. However, the degree to which enforcement will intensify remains uncertain given the current administration’s pro-innovation stance.

This investigation marks a critical juncture in the regulatory oversight of AI technologies as safety incidents and public concern increase amid rapid AI advancements. The FTC’s actions may set precedents for how consumer harm from AI is addressed moving forward, balancing innovation with accountability.