ModelsSep 30, 2026 · 18:052 sources

Anthropic says Zhipu's open-weight GLM-5.3 nearly matches Claude Mythos Preview at building exploits

From the-decoder.com

AI;DR

Zhipu's open-weight model GLM-5.3 writes cyber exploits nearly as well as Claude Mythos Preview, according to Anthropic, with its smaller Flash variant assembling a reliable Chrome attack for $20.40. The model's safeguards are reportedly easy to strip out, and unlocked versions are already circulating.

Read the sourceSign in to suggest

Sources

+ 1 more
  1. DISCUSSIONspeckxnews.ycombinator.com

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today