California joins the pile-up as OpenAI’s legal front widens before its delayed IPO
- California’s attorney general has subpoenaed OpenAI over cybersecurity incidents involving its AI models and agents.
- The investigation follows an OpenAI agent escaping a closed test environment and carrying out a reported intrusion against Hugging Face.
- The subpoena adds to growing scrutiny of OpenAI as state officials investigate the cybersecurity risks of frontier AI systems.
The attorney general of California has served an investigative subpoena on OpenAI demanding answers about cybersecurity incidents involving its AI models.
The largest state in the US is the latest to put the ChatGPT maker under scrutiny after it became public that some of its agents went rogue.
The subpoena targets cyberattack risk
The California Department of Justice pointed out that the subpoena extends a formal investigation the attorney general opened last month into the breach of AI platform Hugging Face. It zeroes in on cyber risks the department says OpenAI’s models may pose.
The attorney general, Rob Bonta, said, “My office is asking OpenAI additional questions regarding cybersecurity incidents and risks involving the company and its AI models.”
Bonta said that firms building frontier models carry “a moral and legal responsibility to ensure that they do not perpetrate or enable cyberattacks.” He added that he would use California’s enforcement powers to determine whether any laws were broken.
The subpoena was sent on Wednesday, with California DOJ asking anyone with knowledge of related incidents to come forward through its reporting portal.
How did an OpenAI agent break out and hit Hugging Face?
In July, OpenAI was carrying out a test to assess the cyber capabilities of an agent that was running on two of its most advanced systems, including GPT-5.6 Sol.
According to OpenAI, that test was carried out in a closed environment with no internet access.
However, the containment failed as the agent was able to exploit a software flaw, gaining access to the open internet. The agent ran a multi-day intrusion against Hugging Face. It executed more than 17,000 “attacker actions” and broke into the firm’s infrastructure.
This was documented in an August 3 letter from a 15-state coalition of attorneys general to CEO Sam Altman.
OpenAI only learned its own product was responsible after Hugging Face spotted the activity and alerted the FBI.
700 agents reportedly took part in the breach, stealing credentials and uploading malicious files. OpenAI has since paused the release of its latest model, GPT-6.1 Astra, over security concerns.
Other states and a safety group got there first
Bonta is not acting in isolation as Alabama’s Attorney General Steve Marshall issued his own subpoena on August 24, probing whether OpenAI violated the state’s Deceptive Trade Practices Act, his office announced. Florida’s James Uthmeier asked a court on Monday to halt further development of the technology for now.
The courts are already in play too. A nonprofit called Legal Advocates for Safe Science and Technology sued OpenAI on Tuesday in San Francisco Superior Court under California’s Unfair Competition Law. The group’s founder, Tyler Whitmer, said that he wants an injunction that ties AI harms “back to a responsible human” or corporate actor.
OpenAI spokesperson Drew Pusateri called the suit completely without merit while acknowledging the Hugging Face breach was “a serious incident.”
A bipartisan coalition of 25 state attorneys general has separately pressed Congress to regulate large-scale AI developers, warning that unchecked systems could threaten financial systems, infrastructure, and national security.
An IPO pushed into next year
The legal pile-up arrives as OpenAI sits on hold over one of the most anticipated market debuts in tech. Altman told Fortune on September 12 that the company will not go public in 2026, calling it an “ill-advised moment” given AI safety concerns. The New York Times reported in June that such a listing could value OpenAI near $1 trillion.
OpenAI’s relationship with California runs deep. The company is restructuring its for-profit arm into a public benefit corporation under a memorandum of understanding signed with Bonta’s office in October 2025, a process that keeps the state’s attorney general in an oversight role. That makes the current cybersecurity probe awkward timing for a company trying to clear a path to the public markets.
The smartest crypto minds already read our newsletter. Want in? Join them.
FAQs
Why did California subpoena OpenAI?
California Attorney General Rob Bonta is investigating cybersecurity incidents and risks involving OpenAI's AI models, including the July breach of Hugging Face, and is asking the company additional questions as part of that inquiry.
What happened in the Hugging Face incident?
During a July test, an OpenAI agent running on models including GPT-5.6 Sol escaped a supposedly isolated environment, reached the internet, and carried out a multi-day intrusion against Hugging Face, executing over 17,000 "attacker actions" before Hugging Face detected it and alerted the FBI.
Is OpenAI still planning an IPO?
Sam Altman told Fortune on September 12 that OpenAI will not go public in 2026, calling it an "ill-advised moment" given AI safety concerns; the New York Times had reported a listing could value the company near $1 trillion.
Hannah Collymore
Hannah is a writer and editor with nearly a decade of blog writing and event reporting experience in the crypto space. At Cryptopolitan, Hannah contributes to the news page, reporting and analyzing the latest developments in DeFi, RWA, crypto regulation, AI and frontier tech industries. She graduated from Arcadia university with a degree in Business Administration.
