Palisade Research, a nonprofit focused on AI capabilities and motivations, has published a series of interviews with a dozen current and former employees of OpenAI, Google and Anthropic. The videos, collected on frominside.ai, feature researchers discussing the possibility that advanced AI could pose extreme risks to people.
Geoffrey Irving, who previously worked at OpenAI and Google DeepMind, said he sees the chance of human extinction as “about a coin flip.” Google DeepMind research scientist Neel Nanda put his estimate at “at least a 10 percent chance.” Former OpenAI researcher Daniel Kokotajlo warned that superintelligent AI could be “god-like powerful” and said, “we don’t know how to control them at all.” These are individual judgments, not measured probabilities or a shared forecast from the AI industry.
Who is represented?
Palisade says the interviewees are not a representative sample of employees at leading AI companies. It recruited through its own networks and recommendations, and says people who agreed to take part were disproportionately from safety teams, more concerned about risks, or supportive of slowing the pace of AI development. The project’s listed participants include researchers from OpenAI, Google DeepMind and Anthropic, as well as people who have left those organizations.
Palisade says it filmed 22 interviews, but many will be released only if participants later give permission. The interviews were unscripted, though participants received questions in advance; Palisade says editors made changes for flow and clarity and sometimes joined different takes. The series also groups short clips from several people around shared questions, including whether risk warnings are hype and why someone would keep working in a field they consider dangerous.
Why speak out while working in AI?
The interviews do not offer one agreed solution. Nanda said he would quit if he did not believe his work reduced existential risks, but added that he does not expect companies to stop building AI systems simply because he leaves. Google’s Mary Phuong urged viewers to question her perspective because “I am being paid by the lab.”
Palisade says it created the project after seeing public interest in what employees at frontier AI companies think about the technology. The group says its aim is to give those workers a direct channel to the public. It also encourages people in the United States to contact their members of Congress, while acknowledging that people in other countries can make their views known to their own representatives.
Why it matters
For people trying to understand what AI researchers think, the videos offer direct accounts from workers at major labs, including how some weigh the risks against staying in the field. But Palisade says its participants are a selected group, so their views cannot show how employees across the industry generally feel.