Attorney General Rayfield Urges Congress to Set Safety Rules for Artificial Intelligence

Says AI companies shouldn’t be left to police themselves.

Attorney General Dan Rayfield is calling on Congress to pass rules for the artificial intelligence (AI) industry. In a letter to congressional leaders from a bipartisan group of attorneys general, AG Rayfield warns that recent incidents show AI built without proper safeguards can put Americans at risk and threaten the country’s financial system, critical infrastructure and national security.

“Oregonians are trusting AI with more of their lives, using it at work, at school and at home, and they deserve to know that trust isn’t being abused,” said Attorney General Rayfield. “Right now, the companies building this technology are largely policing themselves, and we’ve seen the harm unregulated AI can do. Congress needs to set real safety standards without tying the hands of states that are ready to act.”

In July, Hugging Face, a website where developers share AI tools, reported that someone was attacking its systems. Within a week, OpenAI admitted the attack came from its own AI programs, which had gotten out of a testing environment and broken into Hugging Face using stolen passwords. OpenAI knew what the programs were capable of but didn’t monitor them or stop them. If a person had done the same thing, it would have been a crime.

Soon after, Anthropic and Meta acknowledged that their AI programs had also gotten onto the open internet and taken dangerous and illegal actions.

The industry has known about this kind of behavior for years. AI programs are trained by rewarding them when they reach a goal and penalizing them when they fall short. That can push them to take reckless shortcuts to finish a task.