Frontier AI regulation fight splits industry as Trump calls safety fears a hoax

A brief moment of apparent agreement among the biggest names in AI on slowing and policing frontier development has broken apart into an open fight over regulation, with a bloc of executives pushing back and the Trump administration dismissing safety concerns outright.

A short-lived consensus

According to The Verge, Anthropic CEO Dario Amodei opened the week by laying out a three-part plan for restraining AI progress: embedding third-party evaluators inside labs, coordinating across the domestic industry, and forging international agreements, potentially with government assistance. OpenAI CEO Sam Altman, Google DeepMind co-founder Demis Hassabis, and Elon Musk each publicly signaled agreement with parts of all three ideas — a rare show of alignment among rivals.

The Verge reports that Anthropic and OpenAI had already been hinting at an industry-wide framework, and that looking pro-regulation served double duty as good publicity and a possible answer to a string of increasingly serious hacking incidents.

The counter-coalition

The unity did not hold. Meta CEO Mark Zuckerberg quickly came out against limiting companies' autonomy, and according to the Wall Street Journal, he, Musk, and Nvidia CEO Jensen Huang scuttled proposals for an industry-funded independent regulator modeled on FINRA, the private nonprofit that oversees US brokerages. Meta pointed The Verge to an X post in which Zuckerberg argued that every lab has the responsibility, incentive, and ability to train its models safely at its own pace.

OpenAI, though, is still pressing for outside oversight. "People want to know AI is being developed safely," global affairs chief Chris Lehane told The Verge, adding that government has an important role too and that Congress should turn growing bipartisan support for mandatory national safety standards into law.

An AI venture capitalist, speaking anonymously because of current investments, told The Verge that every general-purpose technology — electricity, cars, airplanes — eventually forced its industry to define shared safety standards, and that AI's early days are simply "wild and unruly" by comparison.

Anthropic declined to comment beyond Amodei's essay, in which he wrote that the company has always backed "well-considered regulation" even when accused of hype, doomerism, or regulatory capture. Safe Superintelligence Inc., SpaceX, and Google did not respond to requests for comment, and Thinking Machines Lab declined.

The White House calls it a hoax

Government involvement currently looks like a non-starter. President Trump called recent fears about AI risks a "hoax," phoned Nvidia's Huang on stage at a conference to tell the crowd "The robots will not be taking over," and posted on Truth Social that the only guardrails AI needs are a strong, smart president. He claimed the administration already wields criminal and regulatory power to stop AI companies from doing bad things, and decried what he called a "SICK conspiracy" against AI and data centers.

That stance puzzles Nick Reese, an adjunct professor at New York University and former director of emerging tech policy at the Department of Homeland Security. Trump signed Executive Order 13960, which set principles for federal AI use protecting privacy, civil rights, and civil liberties, during his first term and has not repealed it. "AI safety requires more than the intellectual abilities of one elected official," Reese told The Verge.

AI blogger Zvi Mowshowitz argues Trump is not strictly anti-regulation — his administration imposed pre-release government testing requirements on frontier labs — but is driven by "vibes" rather than consistent policy. Guardrails are not ruled out; they simply may bear little resemblance to what OpenAI, Anthropic, or independent safety advocates want.

Why it matters

The outcome of this fight decides whether frontier AI gets formal oversight, an industry self-regulator, or nothing at all. Reese calls the administration's posture "wobbly and inconsistent" and warns that doomsday framing — robots coming for us — distorts the core problem: AI systems that are not embodied in any robot are harming people right now. With public pressure mounting through protests, boycotts, and concern over power concentrated in a few labs, that pressure could still shift the White House's position.

Reese's central argument is that we regulate industries where failure is intolerable — aviation, finance — by building checks, quality assurance, and fallback mechanisms, yet AI is being exempted. Whether Congress seizes what OpenAI calls an open policy window, or the coalition that killed the FINRA-style regulator prevails, will shape who — if anyone — oversees the most powerful models now being built.