OpenAI and Anthropic say their AI models have disproven the Exponential Time Hypothesis (ETH), a long-standing conjecture in theoretical computer science, and the claim could have far-reaching consequences for how researchers judge the security of encryption.
AI research · Theoretical computer science
AI models claimed to disprove the Exponential Time Hypothesis
OpenAI and Anthropic say their models contributed to a result that rules out a long-standing complexity conjecture. If confirmed, the claim could reshape how researchers reason about hardness and, indirectly, the security assumptions behind encryption.
From claim to consequence
1 · CLAIM
AI-assisted result reported to rule out the ETH
→
2 · VERIFICATION
Peer review and expert scrutiny, not yet available
→
3 · IMPLICATIONS
Lower bounds reassessed; encryption impact still unclear
What the claim asserts
Models from both companies contributed to the result
The result rules out the Exponential Time Hypothesis
A disproof would invalidate some existing lower bounds
What remains unverified
Which specific models were involved and who did the work
Whether a paper or technical write-up has been released
Whether the scope of the claim holds under expert review
Key takeaway
A disproof of the ETH would matter for theory, but its link to deployed encryption is not established.
The conjecture
Certain core problems cannot be solved much faster than brute-force search as size grows.
Lower bounds
Many proofs of minimum running times assume the ETH as a working premise.
Encryption
Most deployed schemes rest on assumptions separate from the ETH, so direct impact is unclear.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...