Checked Oct 1, 2026Start a new check

What was checked

AI just solved a decades old “holy grail” problem in probability theory that resisted elite mathematicians for decades. Anthropic AI generated a proof for percolation theory conjecture in dimensions 3-10 that could plausibly have been Fields Medal level work.

Assessment summary

Needs caution

This is a speculative judgment attributed to a mathematician in media and social-media reports, not an objective fact. The eventual significance cannot be assessed until the proof is independently checked and accepted by experts.

Key findings

  • AI just solved a decades old “holy grail” problem in probability theory that resisted elite mathematicians for decades.

    Supported
    The vanishing of the percolation probability at criticality for dimensions 3–10 was an open problem for many years, and recent reports say an AI-generated formal proof claims to settle these cases.

    Only one AI assessment covered this finding.

  • The problem was a decades-old “holy grail” problem in probability theory.

    Needs context
    The critical-percolation question is a longstanding and important open problem, especially in dimensions 3–10. “Holy grail” is informal promotional language rather than a precise mathematical classification.

    Only one AI assessment covered this finding.

  • The work could plausibly have been Fields Medal-level work if completed by a human.

    Can’t verify
    Whether a result merits a Fields Medal is a subjective, community judgment depending on novelty, depth, and peer acceptance; this cannot be verified from the available reports alone.
  • Anthropic AI generated a proof resolving a long-standing percolation theory conjecture in dimensions 3–10.

    Can’t verify
    Reports describe a Claude-generated Lean development claiming to prove the relevant Kozma–Nitzan conjecture and thereby settle the open dimensions 3–10. However, available reports state that independent mathematical verification has not yet occurred.

    Only one AI assessment covered this finding.

Model assessments

Assessment summary

  • The underlying percolation result and its relevance to dimensions 3–10 are reported by several sources, but the proof is not yet independently validated by referees.
  • The claim that it was definitively solved by Anthropic AI and was Fields Medal-level work is therefore premature and partly subjective.

Claims reviewed

The problem was a decades-old “holy grail” problem in probability theory.

Needs context
The critical-percolation question is a longstanding and important open problem, especially in dimensions 3–10. “Holy grail” is informal promotional language rather than a precise mathematical classification.

This claim may mislead readers without added context.

Anthropic AI generated a proof resolving a long-standing percolation theory conjecture in dimensions 3–10.

Can’t verify
Reports describe a Claude-generated Lean development claiming to prove the relevant Kozma–Nitzan conjecture and thereby settle the open dimensions 3–10. However, available reports state that independent mathematical verification has not yet occurred.

The work could plausibly have been Fields Medal-level work if completed by a human.

Can’t verify
This is a speculative judgment attributed to a mathematician in media and social-media reports, not an objective fact. The eventual significance cannot be assessed until the proof is independently checked and accepted by experts.

Assessment summary

  • Anthropic's AI did release a proof of a percolation theory conjecture, confirming that the transition in dimensions 3-10 is continuous.
  • A mathematician stated that solving this problem would likely earn a Fields Medal.
  • The claim accurately reflects the significance and scope of the achievement, though independent verification is still pending.

Assessment summary

  • Contemporary reporting and repositories show Anthropic's Claude produced a formalized proof claiming θ(p_c)=0 for the previously open dimensions 3–10.
  • However the work has not yet undergone independent human refereeing and acceptance by the mathematics community, and the ‘‘Fields Medal level’’ characterization is subjective and not verifiable.

Claims reviewed

The result could plausibly have been Fields Medal level work.

Can’t verify
Whether a result merits a Fields Medal is a subjective, community judgment depending on novelty, depth, and peer acceptance; this cannot be verified from the available reports alone.

AI just solved a decades old “holy grail” problem in probability theory that resisted elite mathematicians for decades.

Supported
The vanishing of the percolation probability at criticality for dimensions 3–10 was an open problem for many years, and recent reports say an AI-generated formal proof claims to settle these cases.

Anthropic AI generated a proof for percolation theory conjecture in dimensions 3-10.

Supported
Multiple public repositories and news outlets report that Anthropic's Claude produced a Lean/formal artifact claiming a proof that covers the 3–10 dimensional cases.

Sources cited

29 sources

Share this result