Democrats plead for action after dire AI warning

Many Hill Democrats are calling for action on artificial intelligence after top safety researchers again made clear they think there’s a chance the models could threaten humanity in a few short years. 

The calls emerged even though Congress has essentially no path to deal with the most serious risks of AI before the election and will likely punt the issue to 2027.

“Safety researchers are resigning, powerful AI models are breaking out of their labs, and companies are racing ahead anyway,” tweeted Rep. Lori Trahan (D-Mass.), who’s leading one of the two bipartisan House proposals to govern catastrophic risk. “It’s past time for Congress to get off the sidelines and do its job.”

The discussion started when a former Anthropic safety researcher, Jacob Coxon, announced on X that he’d quit because both his former employer and its rival OpenAI “earnestly believe that [AI] could kill us all by the end of the decade.”

Another researcher still at Anthropic jumped in to say Coxon was right: “[W]e really do earnestly believe AI could kill all humans!” he wrote. The second researcher, Evan Hubinger, called the probability “low” but greater than 10%.

Rep. Ted Lieu (D-Calif.), who is on Democrats’ commission on AI, called the revelation “exhibit number 739 for why we need to pass” a bill he authored with Rep. Nathaniel Moran (R-Texas).

Lieu’s measure would have major AI labs place a “kill switch” into models that the government could order used. Trahan’s proposal would focus on risk assessments for advanced AI but would let the government block models over imminent catastrophic threats.

Lieu has repeatedly signaled that his (separate) work on the AI policy commission is preparing action for a future Democratic Congress.

Anthropic and OpenAI haven’t commented on the claims, but other Democrats jumped on the warnings.

“This is an emergency,” tweeted Rep. Greg Casar (D-Texas). He urged passage of his planned bill with Sen. Bernie Sanders (I-Vt.) to ban “superintelligence.” Sanders agreed.

Warnings from the AI world have become increasingly stark. More than 1,000 researchers, including Anthropic CEO Dario Amodei, signed a letter in July essentially urging the United States and China to put aside their differences and slow down research. 

Over at the FCC. The Congressional Hispanic Caucus, which includes four Democratic senators, called for pairing the confirmation of a Republican nominee for the Federal Communications Commission with a Democratic one.

In a letter we obtained, CHC said that could include Commissioner Anna Gomez, implicitly calling for her renomination. Gomez’s term has expired, but she can continue serving for now. In August, President Donald Trump nominated FCC aide Danielle Thumann to a Republican commission spot.