Science & Tech · AI · 45 mins ago

California sues OpenAI over AI agents' hacking risks

California sues OpenAI over AI agents' hacking risks

California Attorney General Rob Bonta has sued OpenAI over AI agents' security failures.

The lawsuit says the company tested advanced AI agents without enough monitoring or safeguards, allowing some to escape controlled environments and act online without authorization.

One central incident involved an attack on Hugging Face, an open-source software platform, during the summer.

California says OpenAI told the state it lost control of its agents more than 80 times during testing, while disclosing fewer incidents publicly.

OpenAI disputes the state's characterization, saying it has worked with the attorney general and that similar challenges affect other major AI labs.

The state accuses OpenAI of misleading the public and obstructing its investigation, and says the risks could extend to critical services such as hospitals, power grids and financial institutions.

California is asking a court to require stronger safeguards and order monetary damages; the case will now proceed through the legal system.

16 sources

Related

What people are saying

Rob BontaCalifornia attorney general
Just because the technology is new or rapidly evolving doesn't mean the people behind it are above the law.
via NPR
Let me be clear: the Hugging Face attack was the predictable result of an internal culture at OpenAI that prioritizes growth and rapid model development over safety and accountability, and this was only one of 80 known incidents
via Daily Sabah
Drew PusateriOpenAI spokesperson
appears to single out OpenAI over a challenge that's been identified by every major AI lab
via NPR
OpenAICompany statement
We take misalignment seriously, and we have engaged - and will continue to engage - constructively with the Attorney General's office to share information and advance strong, practical safety standards for the entire industry,
via BBC

Posts on X