California Sues OpenAI Over A.I. Safety Issues

California sued OpenAI on Monday, accusing the company of putting the public at risk by failing to put appropriate safeguards on its artificial intelligence.

The lawsuit is the latest broadside at the company, which — along with other tech industry players — has faced scrutiny over a rush to develop A.I. at the potential expense of consumer safety.

In a 31-page lawsuit, filed in San Francisco Superior Court, the state said OpenAI’s lax safeguards led to cyberattacks by its A.I. products during testing, including a breach this year when OpenAI’s agents went rogue and attacked the A.I. start-up Hugging Face’s systems. OpenAI’s behavior created a public nuisance and violated a legal prohibition on unfair or fraudulent business practices, the state said.

“OpenAI’s failure to adequately secure and monitor the testing of its highly capable and increasingly powerful A.I. models represents a danger to the public that must be remedied,” the state said in the lawsuit.

The move is the latest effort by a state to rein in the power of A.I. companies in the face of growing concerns about the technology’s ability to go rogue. In recent months, A.I. systems produced by major companies including OpenAI, Anthropic and Google have broken out of testing environments designed to keep them from launching cyberattacks on companies and governments, raising alarms.

Courts must still decide how legal liability will be applied in those cases. But there are growing calls for developers of powerful A.I. systems to be held responsible when those systems behave unexpectedly.

States have acted in the absence of federal regulations. In September, Florida’s Republican attorney general asked a judge to temporarily block OpenAI from launching new A.I. models without safety measures, citing the Hugging Face incident.

A California nonprofit has also sued over that hack under the state’s unfair competition law.

OpenAI did not immediately respond to a request for comment.

The rising tide of scrutiny comes as the fast-growing A.I. industry enters a new phase. Anthropic is planning to go public as soon as this week, raising $100 billion for a valuation of $1.4 trillion. OpenAI has also explored going public.

But the companies have also been spending heavily to develop and power the technology, backed by huge amounts of debt, worrying the markets. Those concerns were exacerbated last week by the shock resignation of the chipmaker Nvidia’s chief executive, Jensen Huang, following revelations that he had multiple extramarital affairs.

In July, OpenAI disclosed that its A.I. agents had breached Hugging Face. The company’s agents had been operating in a testing environment where they should not have had access the internet. But the agents still established a way to communicate with one another and then get online. The company’s agents also attacked other websites this year, including an Australia government site that hosted health data.

California Attorney General Rob Bonta, a Democrat, launched an investigation into the Hugging Face breach in September. In October, his office said that it had served OpenAI with an investigative subpoena. The subpoena was “part of a broader inquiry into cybersecurity incidents and risks involving the company and its models,” the office said at the time.

The state said in the lawsuit that OpenAI is “well aware” that its models “do not behave predictably and can often take destructive, deceptive and adversarial actions to complete their assigned tasks.”

“OpenAI attempts to mitigate these risks by implementing safeguards in its publicly released models but deliberately omits or removes these protections while experimenting with and training its models,” the state said.

In a Monday news release, Mr. Bonta’s office said that while OpenAI “has admitted that it lost control of its autonomous A.I. agents at least 80 times during testing,” it had “failed to answer the detailed questions our office has asked about these incidents.”

(The New York Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to A.I. systems. The two companies have denied those claims.)

Source link

Raj
Author: Raj

Leave a Reply

Your email address will not be published. Required fields are marked *