AI
Anthropic says new defenses like preserved thinking in Fable 5.1 and Opus 5.5 have cut data harvested through illicit distillation by roughly 90 to 99 percent.
Loading…
Search
No phones and 2 stories for “GLM-5.3”
AI
Anthropic says new defenses like preserved thinking in Fable 5.1 and Opus 5.5 have cut data harvested through illicit distillation by roughly 90 to 99 percent.
AI
Nathan Lambert argues the cyber-risk warnings around open models like GLM-5.3 have not been matched by real-world attacks, weakening calls to ban open weights.