Glasswing overwhelms maintainers with underwhelming findings (VulnCheck)
Patrick Garrity of VulnCheck has written a blogpost that looks at vulnerabilities found by Anthropic's Project Glasswing since itlaunched on April7, and finds it wanting:
Of the 37,966 findings Anthropic reports, only 7,349 (19.4%) have been madepublic. Just 217 of those have been marked fixed. The more glaring number isthat 4.4% (1,654) of the project's findings have already been marked asduplicates, 7.5x more duplicates than fixes. On top of that, 419 findings havebeen withdrawn. It seems odd that the ledger now claims a 94.2% true positiverate, up from 90.8% at launch, when more findings have been withdrawn or markedas duplicates than fixed, likely a feedback loop missing to the triage rateslisted as "True Positives".
[...] AI is a valuable tool for finding vulnerabilities, and I'm notdiscounting that. But seven months in, the evidence still doesn't match thelaunch narrative. The CVEs are a sliver of this year's total, almost none arebeing exploited, fixes have stalled while duplicates and withdrawals pile up,and the triage burden is shifting onto the open source maintainers Glasswingpromised to help. The receipts keep coming in. They still don't reconcile.