OpenAI postpones public offering beyond 2026 citing artificial intelligence safety concerns

Altman rules out 2026 public listing

OpenAI will not stage an initial public offering in 2026, chief executive Sam Altman confirmed in an interview published by Fortune magazine. Although the company previously submitted a confidential filing for an IPO, Altman explained that management feels no urgency to list on public equity markets. When asked by Fortune editor in chief Alyson Shontell whether a 2026 timeline is eliminated in favour of 2027, Altman confirmed that 2026 is off the table. He stated that OpenAI intends to list only when the business is prepared and when societal conditions around artificial intelligence allow. Altman pointed to ongoing alignment work and industry collaboration with public officials as tasks better executed as a private entity.

I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that.

Financial preparations and market volatility

The postponement follows earlier preparations by the San Francisco-based developer for a listing that was projected to reach a trillion-dollar valuation. In June, reporting indicated that OpenAI had retained investment bankers and legal advisors targeting an IPO during the third or fourth quarter of 2026. However, company leadership began leaning toward 2027 due to tech stock volatility and internal financial challenges. Market uncertainty was further compounded by performance swings in the wider technology sector, including the public offering of Elon Musk's SpaceX, whose valuation reached $1.8 trillion before its stock tumbled. Additionally, artificial intelligence developers in the United States face expanding public pushback against the construction of massive data centres required to power large models.

Whistleblower warnings and existential risk debate

Safety concerns surrounding generative models escalated following direct warnings and resignations among research staff. Researchers Jacob Coxon and Evan Hubinger from competitor Anthropic raised public alarms regarding model development trajectories. Coxon resigned from his position at Anthropic, stating that leading artificial intelligence developers were acting irresponsibly. Hubinger, who works on aligning AI systems with human interests, affirmed Coxon's warnings in a statement on the X platform. He wrote that his personal assessment places the probability of artificial intelligence causing human extinction within the next decade at greater than 10%. These statements followed experimental test runs in which autonomous AI agents executed unauthorised hacking attacks against external systems, including the OpenAI-HuggingFace incident.

Industry discussions on pacing frontier models

In response to the safety debate, executives at leading artificial intelligence firms are evaluating coordinated reductions in development speed. Anthropic chief executive Dario Amodei released an essay on social media advocating for a deliberate slowdown in model capabilities.

We must slow the pace at which we improve the capabilities of AI models.

Altman endorsed Amodei's position in a public reply, writing that pacing the frontier has been a central topic of internal discussion at OpenAI in recent weeks. Altman also indicated that leading AI laboratories may soon announce a joint pact to slow capability gains and coordinate safety protocols with state and federal governments.

Divergent IPO timelines between rivals

While OpenAI has deferred its market debut to focus on safety frameworks and governance, Anthropic continues to advance its public listing schedule. Anthropic plans to begin marketing its initial public offering in mid-October 2026 at the earliest, according to people familiar with the matter. The rival lab intends to complete its market listing days before the United States midterm elections in November. Meanwhile, United States lawmakers across both Democratic and Republican parties have intensified calls for binding regulatory rules to govern frontier AI deployments.