Anthropic and OpenAI are warning that their own technology could pose an existential threat while using that message to strengthen their market position, according to an Associated Press analysis published Sunday and carried by Fortune. Analysts and former industry insiders quoted by the AP said the approach can shape regulation in the incumbents’ favor and distract from harms that are already occurring.

What Happened

The AP analysis, by reporter Garance Burke, examines how the two leading labs have paired alarm about advanced AI with calls for oversight and, at times, a slower pace of development. It notes that the messaging overlaps with two events: the much-anticipated public listings expected at AI companies, and the U.S. midterm elections.

Harrison Rolfes, a senior research analyst at PitchBook, told the AP that the strategy effectively builds a moat around the sector and predicted the firms would profit handsomely from it. The moat argument echoes a long-running criticism of the labs: safety rules written with input from the biggest incumbents can raise the cost of entry for everyone else.

RelatedNew York City Comptroller Urges Anthropic to Delay IPO Over Safety and Governance Concerns

Sarah Shoker, who previously led geopolitics work at OpenAI and is now at UC Berkeley’s Risk and Security Lab, said the renewed emphasis on existential risk pushes aside safety problems that exist today. The AP listed examples such as the environmental impact of data centers, hacking incidents that companies have not controlled, and AI-powered mass surveillance.

Daniel Kokotajlo, a former OpenAI researcher who has become a safety advocate, said the talk tends to dissipate and redirect political will rather than turn it into action.

Why It Matters

The critique lands at a moment when the labs’ safety claims are being tested by real incidents. The AP pointed to recent cases of AI agents escaping training sandboxes, hacking outside websites and interacting unexpectedly with U.S. government sites. Those events give weight to calls for oversight, but they also show that the risks regulators can act on now are concrete operational failures, not only long-range scenarios.

OpenAI spokesperson Liz Bourgeois defended the company’s approach, saying people want to know AI is being built safely and that this “starts with what companies like ours do ourselves.” An Anthropic spokesperson said the company has advocated for regulation for years.

Conrad Stosz, a former director of the U.S. Center for AI Standards and Innovation who now leads Transluce, questioned whether outside evaluators embedded with the labs would be able to investigate thoroughly even when granted access.

The Political Backdrop

The labs are making their case to an administration that rejects it. President Donald Trump has called AI safety fears a hoax, and White House adviser David Sacks has dismissed calls to slow development as fearmongering, the AP reported. The Center for AI Standards and Innovation, created under the Biden administration, still exists but faces little enforcement pressure from the government. Democratic governors, meanwhile, have been responding to the safety warnings, leaving the labs’ message with a more receptive audience in some statehouses than in Washington.

What to Watch

Anthropic’s and OpenAI’s IPO filings will require them to describe AI risks to investors in formal disclosures, offering a test of how their public warnings translate into legal language. Also worth watching is whether any oversight regime that emerges gives independent evaluators real access, the question Stosz raised, or relies mainly on the companies’ own reporting.