Palantir CTO says some AI researchers started smoking expecting cancer cure

By Splitfeed EditorialLast updated Oct 13, 2026, 03:00 UTC

On October 12, 2026, remarks by Palantir CTO Shyam Sankar drew attention after he claimed that some AI researchers have taken up smoking on the expectation that artificial intelligence will eventually cure cancer. Sankar used the anecdote as an illustration of what he described as extreme optimism inside the field, framing it as a behavioral bet on AI's medical capabilities. He placed this alongside extreme pessimism, calling both ends of the spectrum forms of "social delusion." The remarks were reported by outlets including Business Insider, Benzinga, and Yahoo Finance.

The significance of Sankar's comments lies less in the smoking anecdote itself and more in what it suggests about the cultural mood among some AI researchers. If taken at face value, it implies that confidence in AI-driven medical breakthroughs has moved beyond professional enthusiasm into something that shapes personal risk decisions. Sankar's choice to criticize both poles of AI sentiment is also notable given his employer. Palantir is a major commercial beneficiary of AI adoption, so a senior executive cautioning against uncritical optimism carries a different weight than the same argument made by an outside critic. What remains genuinely in dispute is whether the smoking claim reflects a real and meaningful pattern or is a rhetorical flourish, and whether Sankar's framing of symmetrical delusion is itself a balanced position or a way of softening a critique of hype.

None of the three camps had published reactions at the time of this report, so what follows reflects what each would typically argue in a story of this kind. Pro-AI voices would likely defend the underlying logic, arguing that if AI timelines for medical breakthroughs are credible, recalibrating personal risk behavior in response is rational rather than delusional. Anti-AI voices would probably treat the anecdote as confirmation that AI hype has reached a dangerous and absurd extreme, one that is now influencing real-world behavior in harmful ways. The middle ground camp would most likely agree with Sankar's framing, seeing the smoking story as a cautionary example of how both overcorrection and undercorrection distort reasonable assessment of what AI can realistically deliver.

The argument would sharpen considerably if Sankar or another Palantir figure elaborated on the claim with more specifics, or if researchers who fit the description chose to respond publicly. Any credible evidence from oncology or clinical AI research over the next year showing either a breakthrough or a plateau would also reframe how the anecdote is read in retrospect.