6 ms·
What's interesting to me about this is that it targets Claude's specific tics. Anthropic has created model that reliably reaches for the same tools (yes, and ph
by colinmarc 17d ago
What's interesting to me about this is that it targets Claude's specific tics. Anthropic has created model that reliably reaches for the same tools (yes, and phrases; `python -c` is a load-bearing tool for it). Everyone gets the same model, so by learning the model's behavioral patterns you can target it better.
- throwawayffffas 17d agoKimi and GLM reliably run python to do stuff as well.
- andai 17d agoIf my memory serves me so do GPT and Deepseek. So I'm not sure if this attack is Claude specific at all.