Artificial Intelligence & Machine Learning
    ,
    Next-Generation Technologies & Secure Development

    Altman Says Real-World Deployment Remains Essential to Understanding AI Risks

    Emilia David •
    November 18, 2026    

    Altman Defends OpenAI's Approach to AI Agent Safety
    OpenAI CEO Sam Altman on Sept. 25, 2026 during a public appearance in Kolkata, India. (Image: Shutterstock)

    OpenAI CEO Sam Altman said that while he wishes the Hugging Face incident hadn’t happened, the company had no way of knowing how unsafe the agents were.

    See Also: Webinar | Closing Visibility Gaps. How Leading Organizations Are Securing and Scaling Their Cloud-First Architecture

    Speaking at the New York Times’ DealBook Summit, Altman and OpenAI president Greg Brockman spoke about how the company is navigating what they see as a “newly challenging” period.

    Altman said the company has always followed an iterative deployment approach, releasing new models to smaller groups and adjusting things before allowing wider access.

    “We wish Hugging Face hadn’t happened,” Altman said. “The only way that we have known how to make this technology safe is to confront reality.”

    OpenAI agents coordinated to hack into Hugging Face’s model repository system. The incident fueled the current AI safety debates and put focus on agent behavior. Both OpenAI and Anthropic paused training of their newer models so they could upgrade their safety guardrails and misalignment evaluations.

    Altman is one of the AI executives who expressed support and urged the industry to “pace” AI model development. Slowing down AI innovation has become a controversial topic in the AI world, with some leaders arguing that slowing down will erode the United States’ competitive advantage.

    “We’re not going to make everybody happy,” Altman said. “If we talk too much about taking extra care for safety or there are risks that we really need coordination to mitigate, people are going to accuse us of regulatory capture or trying to slow down innovation.”

    He added that if OpenAI thinks the opposite, “people will say don’t kill us all, a very reasonable thing.” Altman said the company will continue to do what it believes “will actually deliver a safe future with people in control.”

    Comments are closed.