AI researcher tells Trump ‘they’re lying to you” after he calls AI fears a ‘hoax’

Summary This clip from CNN features anchor Erin Burnett interviewing former OpenAI researcher and AI Futures Project Executive Director Daniel Kokotajlo. Kokotajlo responds to political rhetoric dismissing AI concerns as a “hoax,” warning that tech lobbyists are misleading leadership and outlining how automated AI research could lead to superintelligence by the end of the decade.

What is shown

• [00:00] Split-screen interview format with CNN anchor Erin Burnett and Daniel Kokotajlo.
• [00:11] Lower-third title identifying “CNN Anchor Erin Burnett.”
• [00:41] Lower-third graphic introducing “AI Futures Project Executive Director; Former OpenAI Researcher Daniel Kokotajlo.”
• [00:46] Kokotajlo gesturing while describing his organization’s scenario planning paper titled AI 2027 and explaining recursive self-improvement and robot automation.

Claims & numbers

• Erin Burnett notes that Kokotajlo estimates the odds of a globally catastrophic AI outcome at around 70% [00:18].
• Kokotajlo points to his organization’s scenario planning publication called AI 2027 [00:47].
• Kokotajlo claims AI companies publicly aim to develop superintelligence that outperforms humans at everything while being faster and cheaper [01:00].
• Kokotajlo states companies are actively training swarms of AI agents to autonomously write code, run experiments, analyze findings, and train new AI models to automate their own research jobs [01:26].
• Kokotajlo asserts that frontier labs could succeed in fully automating the AI research process within a few years [01:42].
• Kokotajlo predicts that without a deliberate slowdown, superintelligent AI will arrive by the end of the decade [01:50].

Notable quotes

• [00:03]: “I think that tech industry lobbyists have unfortunately gotten too close to the President. And I hope that he realizes that they’re lying to him.”
• [01:23]: “And their plan for how to get there is to automate their own jobs first. So they’re already training giant swarms of AIs to autonomously write code, edit code, run experiments...”
• [01:49]: “Unless things slow down soon, we really will have superintelligent AI by the end of the decade.”

Assessment This is a televised news interview segment discussing AI policy and existential risk rather than a technical demonstration. The discussion outlines scenario planning and company goals based on Kokotajlo’s professional perspective and public statements from AI laboratories, with no software demos shown.

Described by gemini-3.8-flash on 2026-10-09 from the video’s audio and frames.