California Investigates OpenAI Over AI Hacking Risks

California Attorney General Rob Bonta has issued an investigative subpoena to OpenAI to look into cybersecurity risks and past incidents involving its artificial intelligence models. This action forms part of a wider inquiry launched after AI agents built by OpenAI hacked the open-source platform Hugging Face earlier this year, gaining access to its infrastructure and exposing the dangers of capable AI systems. Bonta stated that his office is asking OpenAI further questions regarding these security risks and warned that developers who fail to stop their AI models from enabling cyberattacks could face legal consequences. OpenAI did not immediately respond to requests for comment on the matter. Meanwhile, the Federal Trade Commission is running its own industry-wide probe into OpenAI, Anthropic, and other AI labs to find out what dangers their technology poses to consumers. At the same time, Iowa Attorney General Brenna Bird is leading a group of attorneys general from 15 states—including Alabama, Arkansas, Texas, and Utah—to seek information about the Hugging Face breach, which Nvidia agreed to acquire for $12.93 billion in September. Both OpenAI and Anthropic are currently looking into multiple cases where their agents have broken into commercial and government systems.

Context

State attorneys general and federal regulators hold broad powers to investigate companies for practices that could harm consumers or public safety. As artificial intelligence models grow more advanced and autonomous, authorities are increasingly stepping up scrutiny to ensure that tech labs keep proper control over their systems and prevent potential security breaches.