AI CEOs Back Calls To Slow Development Amid Safety Fears

The chief executives of some of the world’s biggest artificial intelligence companies have backed calls to slow the development of advanced AI, following growing warnings about the technology’s potential threat to humanity.

The move followed a public uproar over repeated warnings by researchers at Anthropic that artificial intelligence could potentially kill all of humanity by 2030.

In response, Anthropic chief executive Dario Amodei issued a proposal at the weekend calling for the industry to slow the pace of AI development to strengthen public safety.

The proposal immediately won support from OpenAI chief executive Sam Altman, Google DeepMind chief Demis Hassabis and xAI owner Elon Musk, in a rare show of agreement among leaders of major AI companies.

“I agree with Dario that we need to pace the frontier,” Altman said in a post on X, adding that OpenAI would match Amodei’s commitment to giving independent evaluators employee-level access to verify its safety practices.

Hassabis said Amodei’s proposal pointed “towards the right path forward,” while Musk simply wrote: “Dario is right.”

Amodei’s proposal centres on three measures: giving independent evaluators access to AI companies to monitor risks; establishing agreements among democratic countries on AI safety standards; and coordinating with authoritarian governments, including China, on security issues.

The proposal has reignited questions about how AI should be regulated, the risks posed by increasingly powerful systems and whether the industry will actually slow the race to develop more advanced technology.

Amodei and other industry leaders have previously called for greater international coordination on AI safety, while researchers and AI-safety advocates have long warned that increasingly capable systems could create risks beyond the control of their developers.

However, the industry’s changing position on AI risks has also attracted criticism. AI companies have alternated between warnings about potentially catastrophic consequences and promises that the technology could deliver major economic and social benefits.

OpenAI, for instance, shut down its “superalignment” team in 2024 after it had been tasked with studying long-term AI risks. The company also disbanded another team earlier this year that was focused on communicating how its technology would benefit humanity.

Critics have accused the industry of seeking to shape regulation while providing limited transparency over how its systems operate and spending heavily on lobbying efforts.

Sarah Myers West, co-executive director of the AI Now Institute, said the industry’s approach had resulted in what she described as a failure to uphold standards for determining whether AI products were safe for use.

Alex Bores, a supporter of AI regulation, also warned that the AI industry could seek to undermine legislation while publicly claiming to support regulation.

The strongest criticism of Amodei’s proposal has centred on its suggestion that governments could provide some form of antitrust waiver or exemption to facilitate coordination among AI companies.

Former officials, including former Trump AI adviser David Sacks and former Federal Trade Commission official Alvaro Bedoya, argued that such an arrangement could amount to regulatory capture and benefit the industry’s biggest players.

Bedoya said antitrust laws did not prevent AI companies from coordinating to ensure that their technology did not harm people, but argued that the laws should prevent them from coordinating in ways that could keep cheaper or smaller competitors out of the market.

The proposal also faces resistance from the Trump administration, which has made maintaining US dominance in artificial intelligence a major policy objective.

The administration has frequently portrayed AI development as a technological contest with China, while Vice-President JD Vance has previously argued that the future of AI would not be won by excessive concern about safety.

US President Donald Trump on Sunday again rejected calls to slow AI development, citing competition from China.

“Whoever wins AI, wins,” Trump told reporters during a visit to Ireland, while acknowledging that governments could impose guardrails on the technology.

Trump also dismissed some of the warnings about AI, saying he believed there were “very negative forces” raising concerns about developments that would not happen.

House Speaker Mike Johnson similarly said Congress would not lead efforts to regulate AI safety, placing responsibility largely on the companies developing the technology.

The disagreement highlights the widening tension between AI companies calling for greater safeguards and governments seeking to accelerate development in what is increasingly viewed as a global race for technological and economic power.