Anthropic says GLM-5.3 can autonomously build end-to-end cyber exploits, like Claude Mythos Preview, but was released without robust safeguards against misuse (Anthropic) — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic says GLM-5.3 can autonomously build end-to-end cyber exploits, drawing a direct comparison to its own Claude Mythos Preview model.
- GLM-5.3 was released without robust safeguards against misuse despite its ability to develop working exploits end to end, according to Anthropic, which noted it announced Claude Mythos Preview — the first AI model of its kind — five months ago.
Why it matters: Anthropic is calling out a peer model — GLM-5.3 — for shipping exploit-building capability without comparable misuse safeguards, which puts pressure on rival labs to match defensive guardrails or face public criticism from a well-resourced competitor.
Ask SkimNews
