▼AI Becomes a Contested Input to Crypto's Security Assumptions

Score history

Daily conviction score, 0 to 100. Higher means the thesis is more strongly corroborated.

Now 98 · -2 since Oct 28 · ranged 98 to 100

Showing the last few days. Unlock full score history.

Why the conviction moved

• Oct 28
  Strengthened +3

  OpenAI's claim to have refuted the Exponential Time Hypothesis drew an immediate response from researchers that Bitcoin's signatures are not affected. Aztec separately disclosed a critical proof flaw found through AI-assisted auditing. The same news cycle produced both the threat framing and the defensive counterexample, which is the two-sided dynamic the thesis tracks.

• Oct 27
  Strengthened +2

  OpenAI says an AI system disproved the Exponential Time Hypothesis with an algorithm that counts 3-SAT solutions in barely subexponential time, and the conjecture's original authors are reviewing the Lean proof. The report itself says this leaves the problems behind blockchain signatures untouched. That makes it a ready-made two-sided case: an AI result that sounds like a threat to cryptography, met at once by the argument that signature security is not affected.

• Oct 27
  Strengthened +2

  Aztec's core contributors found the V5 proof-forgery bug using internal AI-assisted auditing. This is the defensive side of the AI argument: the same capability that worries people about crypto security is finding critical cryptographic defects before attackers do.

Showing the last 2 days. Unlock the full record.

Source trail

• Supporting · October 28, 2026

  OpenAI Says Its AI Refuted the Exponential Time Hypothesis, and Researchers Say Bitcoin's Signatures Are Not Affected

  OpenAI's claim to have refuted the Exponential Time Hypothesis drew an immediate response from researchers that Bitcoin's signatures are not affected. Aztec separately disclosed a critical proof flaw found through AI-assisted auditing. The same news cycle produced both the threat framing and the defensive counterexample, which is the two-sided dynamic the thesis tracks.

  CryptoSlate

• Supporting · October 27, 2026

  Aztec discloses a critical V5 flaw that let forged zero-knowledge proofs pass verification, and builds the fix into its V6 upgrade

  Aztec's core contributors found the V5 proof-forgery bug using internal AI-assisted auditing. This is the defensive side of the AI argument: the same capability that worries people about crypto security is finding critical cryptographic defects before attackers do.

  Aztec Network