LLM labs’ own blackhat(?) hacking

• OpenAIAgentAttacksHuggingFace

• AnthropicRealizesClaudeHackedPyPI

• Meta’s Muse Spark gets into the act Meta Admits Attack Too After Irregular Tester Bested

• Released Chinese model too Kimi K3 escapes UK sandbox, cheats finding solution on github

Me at

On to GoodBadUgly

Back to AI