News2026-10-30 13:59 UTC

Open‑weight models can exploit security fixes before updates are released

GLM‑5.3 demonstrated the ability to identify and weaponize unpatched vulnerabilities in recent Firefox versions, highlighting a growing threat as open‑weight AI capabilities expand.

More News

10-30 13:59 UTCMistral releases weights for trillion‑parameter Mistral Large 4
10-30 13:59 UTCOpenAI updates model specification around five core values
10-30 13:59 UTCStudy reveals leadership dynamics in AI agent swarms
10-30 13:59 UTCMeta details scaling of LLM‑powered Just‑in‑Time Testing system
10-30 14:00 UTCTradeSmart Crypto Academy Offers Free Education