Anthropic has opened an opt-in service that sends projects vulnerability reports generated by its strongest AI models, including Claude Mythos. Crypto projects are already applying, Cointelegraph reports.
The demand makes sense. On-chain code holds user funds directly, and a bug that slips past review can be costly once it ships. A frontier model scanning a codebase before launch gives teams one more pass at catching what humans missed.
I'd still read those reports as leads to check. Before anyone trusts this over a human audit, I want to know how many of the flagged issues turn out to be real. If that number holds up, I expect an AI scan to become a routine step before mainnet.
$OGN $STRK $CTSI
The demand makes sense. On-chain code holds user funds directly, and a bug that slips past review can be costly once it ships. A frontier model scanning a codebase before launch gives teams one more pass at catching what humans missed.
I'd still read those reports as leads to check. Before anyone trusts this over a human audit, I want to know how many of the flagged issues turn out to be real. If that number holds up, I expect an AI scan to become a routine step before mainnet.
$OGN $STRK $CTSI