The OpenAI IPO delay on AI safety grounds is now official: chief executive Sam Altman has said a public market debut this year would be ‘ill-advised,’ pushing one of the most-anticipated listings in recent memory to at least 2027. The announcement arrived on the same day that Anthropic chief executive Dario Amodei published a roughly 3,800-word essay calling on AI companies to slow the pace at which they improve their most advanced models, and Altman and Elon Musk both publicly backed the proposal.
The timing is uncomfortable for OpenAI’s own finances. Chief financial officer Sara Friar had told employees last month that the company would likely go public in 2027 or even sooner, if ‘our business continues to inflect.’ Altman’s about-turn on timing, framed explicitly around safety, is the clearest signal yet that concern over increasingly powerful AI systems is beginning to collide with the industry’s commercial plans.
Amodei’s ‘We Must Pace the Frontier’ essay sets out a three-step framework
Amodei’s essay, titled ‘We Must Pace the Frontier’ and published Saturday, runs to roughly 3,800 words and proposes a structured deceleration rather than an outright halt. The first step is one Anthropic has already taken unilaterally: granting third-party evaluators permanent, employee-level access to verify the company’s safety practices and report incidents. According to Winzheng, the AI safety assessment organisation METR is among the evaluators that will receive that standing access.
The second step calls on leading AI companies within democratic countries to coordinate and establish common safety standards. The third extends that coordination to democratic and authoritarian governments alike. Amodei conceded that some steps may be easier to achieve than others, and he was careful to define the limits of what ‘pacing’ means. ‘Pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models, and for third party evaluators to confirm this,’ he wrote.
Amodei said that while proposals to pause or slow AI development had been circulating since 2023, it made ‘little sense’ to act at that point, because models were not yet powerful enough to take action in the real world and were not yet capable of ‘significant deception, manipulation, cheating, or cyberattacks.’ The situation, he argued, has changed.
OpenAI IPO delay reflects a broader shift in how the industry frames risk
OpenAI‘s chief scientist Jakub Pachocki had earlier this month published a blog post warning that no AI company has ‘solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer.’ Pachocki said he expects voluntary slowdowns to become ‘commonplace until shared safety bars are established.’ In the AI industry, alignment refers to the work developers undertake to ensure systems behave in accordance with human values and intentions.
Altman said on X that the need to pace advanced AI capabilities had been a ‘primary topic’ of discussion at OpenAI in recent weeks. He backed Anthropic’s third-party access commitment directly: ‘Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We’ll have more to share soon.’
Musk, whose AI startup xAI was acquired by his rocket company SpaceX earlier this year, added a blunt endorsement: ‘Dario is right.’ The alignment among three figures who have frequently been at odds with one another marks a shift in tone across the industry’s most prominent voices. Musk had previously been a vocal critic of Anthropic, but his position changed after Anthropic announced a major compute deal with SpaceX in May.
The essay also drew a response from Anthropic’s head of public policy, Sarah Heck, who called on lawmakers to act. ‘The government has a critical role to play here, including blocking the sale of the most advanced chips to adversarial nations like China, enacting a national law requiring testing of frontier models, with the power to block the most advanced models that prove to be unsafe,’ she wrote on X.
Pressure from Congress is already building. Lawmakers in both parties are calling for new AI safeguards and demanding tech leaders testify following a series of cyberattacks carried out without direct human control. With midterm elections approaching, AI has moved firmly onto the political agenda.
Amodei, speaking in a CNN interview after the essay published, drew the boundary around deceleration carefully. ‘If we go too slow, I still believe that the wrong people will be in charge of the technology. And that, again, will bring the probability of things going wrong very high,’ he said. The next concrete step will be Altman’s promised further disclosure on how OpenAI intends to implement independent, employee-level evaluator access.


