Bloomberg report some kids on Discord have had access to Mythos for several weeks due to Anthropic’s shit cybersecurity https://www.bloomberg.com/news/articles/2026-04-21/anthropic-s-mythos-model-is-being-accessed-by-unauthorized-users

[theorangetheme@en.osm.town]

@GossiTheDog Lol. Lmao, even.

[iyashikei_kris@mstdn.social]

@GossiTheDog

Can foresee tomorrow's headlines: "some kids on Discord achieve AGI, use it to make Roblox real"

[thekeystonecollective@infosec.exchange]

Beta-testing it with Discord teens. Truly bringing “discord” back to its original meaning. Chaos, conflict, and the end of tomorrow’s world, one leaked model at a time.

[captmorgan@freeradical.zone]

RE: https://cyberplace.social/@GossiTheDog/116445070012166126

@GossiTheDog This is really bad. Lets hope they weren’t using it to find CVEs in everything

[lycansubscribe@furry.engineer]

@GossiTheDog damn I wonder why mythos didn't just fix the security holes

[tunafishtiger@mastodon.online]

@iyashikei_kris @GossiTheDog @Migueldeicaza

[dcolborne@techhub.social]

@GossiTheDog Maybe Anthropic should have used Mythos against itself before deploying it to production.

[supermoosie@mastodon.au]

@dcolborne @GossiTheDog

Probably relied on it, hence the lack of security

[f4grx@chaos.social]

@lycansubscribe @GossiTheDog it's too good, we're afraid to deploy it internally.

[stuwares@infosec.exchange]

@GossiTheDog The Anthropic quote in the Guardian got me. New euphemism for "we got hacked"