The White House says Anthropic is fearmongering. Anthropic says Washington’s asleep at the wheel.
At the center is a clash of philosophies: how much fear is “appropriate” when building technology that could transform — or destabilize — the world. The dispute came to public attention when Anthropic co-founder and head of policy, Jack Clark, published an essay arguing that policymakers are underestimating AI’s existential risks.
The essay argued that many people are “pretending that AI cannot threaten humanity.” It urged acknowledgment of a “different reality” before the world can determine how to “tame it and live together.”
The White House’s AI adviser, David Sacks, hit back on X, accusing the company of using safety concerns as a business strategy, stating, “Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering.”
Clark, who is a former technology journalist, found Sacks’ criticism “perplexing,” according to a Bloomberg report. He asserted in an interview with the outlet that in “many areas we’re extremely lined up with the admin.” Still, he added that there are “some areas where we have a slightly different view, and we articulate that view in a substantive, fact-forward way.”
He also questioned the lack of policy proposals from the administration, asking whether Sacks had “published an approach for what federal regulation of AI could look like? Because he’s saying that that’s what’s needed.”
The California catalyst and state vs federal rules
While both the company and the administration officially support some form of federal policy, their disagreement over state-level regulations has been a major source of tension.
The White House had previously thrown its weight behind a failed attempt to impose a ten-year hold on state-level AI legislation, arguing that a “patchwork of state regulations” would “sow chaos and slow innovation.”
Anthropic called that proposed moratorium “too blunt.” Furthermore, the company notably stood apart from its industry peers by publicly endorsing California Senate Bill 53, a landmark piece of legislation that imposes first-of-its-kind transparency rules and whistleblower protections on frontier AI developers.
This support for California’s legislation appears to have been a key “catalyst for Sacks’ outburst,” as noted in Bloomberg’s coverage. Anthropic defended the move, with Clark reiterating that their preference was for a federal solution, but saying that since the federal government had failed to “get its act together,” the company had to act.
Axios highlights the overall picture, noting that the fight is as much about state-level rules as it is about federal ones.
Anthropic’s rules frustrate federal agencies
The friction extends beyond legislation and into the daily work of the government. According to a report from Semafor, Anthropic’s strict usage policies have created headaches for federal law enforcement.
The company refuses to allow its AI models to be used for “domestic surveillance,” a policy that has led it to decline requests from contractors working with agencies like the FBI and Secret Service. Senior White House officials, speaking anonymously, told Semafor they believe Anthropic is making a “moral judgment” about law enforcement work.
This creates a practical problem. In some cases, Anthropic’s Claude models are the only top-tier AI systems cleared for top-secret situations through the Amazon Web Services GovCloud system.
This leaves contractors and agencies in a bind, with access to a powerful tool but being blocked from using it for specific investigative purposes.
This ongoing political friction, combined with the fact that Anthropic CEO Dario Amodei publicly supported Kamala Harris and has been “absent from White House tech events,” according to Axios, places the company in an increasingly difficult political spot with the current administration.
Speaking of AI regulation: In June, US senators agreed to try to amend a bill that would bar states from regulating AI, proposing to shorten the moratorium from 10 years to five.
The post Anthropic Feud With the White House Exposes Deep Divide on AI Oversight appeared first on eWEEK.
No Responses