"five months into Project Glasswing, only 9.8% of the findings have made it to a project maintainer"
2 Boosts
0 Likes
number of surprised Open Source maintainers: zero
“The number of vulnerabilities we've disclosed is a subset of the total number of vulnerabilities that Mythos Preview (and other Claude models) has found, since the process of independent human triage and review is the rate limiting step.” —Anthropic
No, Anthropic, the rate-limiting step is that 99% of everything coming out of your damn bots is nonsense, and extracting any shred of useful information from their output is the mother of all needle-in-a-haystack problems.
@argv_minus_one @bagder So essentially they're trying to push the idea that code review and general maintenence should be done by AI. At this point the idea is that a AI write the code and then an AI review the code
@eletrotupi @argv_minus_one I think rather that the marketing was all that noise back in spring and now they don't care anymore...
@bagder Now wait. They've disclosed roughly 2500 findings to maintainers and only roughly 500 have been handled? What happened to the rest of them? Ignored? Laughed at? Reported to projects that were abandoned?