Prompt Injection as Role Confusion
Research reveals LLMs rely on text style rather than tags to distinguish instructions; destyling drops injection success rates from 61% to 10%.
Simon Willison · Jun 23, 2026
Tag: 模型对齐 (2 articles)
Anthropic Walks Back Policy That Could Have ‘Sabotaged’ AI Researchers Using Claude
Anthropic reverses its controversial policy of silently limiting Claude for frontier LLM research, sparking industry-wide reflection on AI safety transparency and developer trust.
Simon Willison · Jun 11, 2026