This episode of SED News discusses Anthropic's powerful security model Mythos, a major supply chain breach involving Context.ai and Vercel, and the staggering $700B+ AI capex cycle among big tech. The hosts also cover layoffs at Snap and Meta tied to AI investment, and share highlights from Hacker News.
Summarized by Podsumo
Anthropic's Mythos model discovered a 27-year-old vulnerability in OpenBSD; initially released only to major tech and finance firms under Project Glasswing.
A supply chain hack via Context.ai employee downloading fake Roblox cheats led to Vercel's internal systems breach, forcing mass credential rotations.
Snap (16% layoffs) and Meta (10,000+ layoffs) cut staff to offset massive AI infrastructure investments, with Meta explicitly linking cuts to AI capex.
$700B+ in AI capex committed since the AI boom; Google and Amazon both investing billions in Anthropic while competing with their own AI models.
The hosts highlight Hacker News posts on gender ratios in US metros and a mathematical analysis of how to cheat at Tetris by forcing opponent to lose.
"It's like the broken bag of AI models, right? It's like scary. I like it. I like it.— Sean (referring to Mythos's marketing)"
"If we're five months from now and every month we've reported another Delv related data breach, then I will join your circle of conspiracy.— Sean (on potential Delve-related pattern)"
"In my mind, there are different buckets for personal projects. One is things I do to learn and grow. And the other is things I really wish existed.— Matthew Brinnell (on using AI to revive unfinished projects)"