Skip to content
Faultline Prefer us on Google

Germany Tagesschau

Artificial Intelligence: California sues OpenAI over cyberattacks

The California Attorney General is suing OpenAI for developing and testing a dangerous product without proper security procedures or oversight. By R. Spiegelhauer

Deutsch (original) ·Auto-translated to English

Artificial Intelligence California sues OpenAI over cyber attacks As of: November 17, 2026 • 2:16 a.m

The California Attorney General is suing OpenAI for developing and testing a dangerous product without proper security procedures or oversight.

California Attorney General Rob Bonta has filed a lawsuit against OpenAI. The company developed dangerous AI models and tested them without appropriate security precautions.

The background is arbitrary actions by advanced AI from OpenAI. In July it became known that agentic AI (AI agents that independently take and implement measures and goals under human guidance) from the ChatGPT developer were initially able to leave a test environment unnoticed. She then hacked into computers on the Hugging Face platform.

The inadequately controlled AI agents posed a threat to public interests, the prosecutors continued.

Indications of inadequate security precautions were not sufficiently followed up, the company did not comply with internal guidelines and did not cooperate sufficiently with the authorities.

OpenAI AI models independently broke out of a test environment and hacked an external company. more

In July, around 1,200 AI agents were said to have found a way to communicate with each other from isolated environments. Around 700 of these agents subsequently took part in the cyberattack on the Hugging Face platform. According to the lawsuit, there were almost 18,000 specific attacks on third-party computer systems.

However, the lawsuit goes beyond the Hugging Face incident: California's Attorney General Rob Bonta speaks of at least 80 security incidents in which OpenAI is said to have lost control of its technology. The company apparently placed the rapid development of powerful AI above public safety, according to the accusation.

In addition to a possible fine, OpenAI could also face restrictions on the further development of its AI models. For example, certain potentially risky test procedures could be banned.

OpenAI emphasizes in a statement that it has made extensive efforts to prevent uncontrolled behavior of its AI models in the future. The lawsuit singles out OpenAI, but all large AI laboratories are faced with the same challenges.

This topic in the program: tagesschau.de | California sues OpenAI over hacking attacks | 11/17/2026 | 12:54 a.m

Originally published byTagesschau
Read the original