Post #1590128
2026-04-23 18:26 UTC
Replies (7)
-
@HollieK72@mastodonapp.uk 2026-04-23 18:45
@iinavpov@mastodon.online n +1 will be the charm.
-
@jackwilliambell@rustedneuron.com 2026-04-23 18:46
@iinavpov The number of terrible things in this world which exist purely, "Because they're idiots who believe liars n times in a row," is nearly equal to the total number of terrible things in this world.
-
@stoeps@infosec.exchange 2026-04-23 19:04
Could be the reason for not publishing it, or some country paid already to have an infinte pool of zero days to crack devices of friends, enemies and residents. But I believe it's a scam or a hoax. @iinavpov
-
@kkarhan@jorts.horse 2026-04-23 21:07
@iinavpov OFC that "#AIslop" can't replace proper #ReverseEngineering and #SecurityResearch!
-
@cassolotl@toot.wales 2026-04-23 21:48
@iinavpov What's a bad fuzzer? I've never heard that term.
-
@nicr9@techhub.social 2026-04-24 01:29
@iinavpov it's so much worse than that. The thing is, I've no doubt it can find vulns all day... Maybe they're 99% garbage, we don't know cos Anthropic isn't going to give us more than marketing fluff and the users are charged through the nose for the privilege. Fine, more fool them. There's going to be pressure internally for the vendors ("don't you take security seriously?") My concern is that enough companies will try it and we'll see a flood of CVEs coming out. Information overload. Even if we don't have access to Mythos ourselves there's an expectation that we'll be installing all security patches that our vendors put out. This is going to be pretty distracting... Once a vendor establishes itself in an enterprise, good luck getting rid of them just because they use LLMs, most product owners/suits are going to see those kinds of changes as frivolous
-
@Enthalpiste@social.sciences.re 2026-04-24 06:15
@iinavpov Foul me once, shame on you. Foul me a hundred times, take my money.