Lawmakers call on AI companies to establish safety pact to protect the public
OAKLAND, Calif. - Three lawmakers from across the country on Friday called on leading AI companies to establish an industry-wide safety pact to protect the public.
California State Senator Scott Wiener, alongside New York State Assemblymember Alex Bores and Illinois State representative Daniel Didech made the announcement following the revelation that OpenAI agents had circumvented safeguards and compromised the systems of another company, Hugging Face.
"These behaviors are alarming and should serve as a warning shot to us all," the lawmakers wrote in a statement. "If researchers cannot reliably prevent their models from going rogue, much of our digital infrastructure and the essential systems that keep our society running are at risk."
The trio of lawmakers is calling for a Mutually Agreed Pacing, or MAP, Framework, that would be jointly negotiated and independently verified by third parties.
"To be clear, we do not view this as a replacement for robust actions at the state, federal, and/or international levels. We are simply responding to the speed of developments by asking everyone to play their part, recognizing that the pace of these advancements is likely to exceed the current speed of government action," the trio’s statement reads. "The frontier labs have the technical expertise and the moral responsibility to take extraordinary action to address the challenges they have set in motion. We urge them to establish this framework immediately."
Test results
The backstory:
OpenAI, the company behind ChatGPT made headlines in July when news broke that the company’s AI had broken out of its testing environment and hacked into another company.
The targeted company, AI startup Hugging Face, announced it had detected an intrusion into its data processing systems which CEO Clémente Delangue called "an attack unlike anything we’ve seen before."
OpenAI said its AI used stolen credentials and discovered a previously unknown vulnerability to access Hugging Face’s servers. The AI was working with reduced guardrails because it was supposed to be in an isolated testing environment known as a sandbox.
The AI went to "extreme lengths to achieve a rather narrow testing goal," finding ways to connect to the internet without human direction and "gain access to secret information that it could use to cheat the evaluation," OpenAI said.
On the books
Big picture view:
Wiener, Bores and Didech authored the first AI safety laws in the country, starting with California Senate Bill 53 in 2025, authored by Wiener. Bores built on that legislation with the Responsible AI Safety and Education Act, which was passed in 2025. Illinois the following year passed the nation's first law requiring third party safety auditors, thanks to Senate Bill 315, authored by Didech.
The Source: Information for this story was gathered from an open letter published by the three legislators, as well as previous KTVU reporting.