Quoting Boris Cherny

1 · · July 25, 2026, 1:02 a.m.
prompt injection Anthropic claude Generative AI prompt injection Generative AI AI LLMs
Summary
Boris Cherny discusses the robustness of the Opus 5 model against prompt injections, noting it as the least susceptible to such attacks in the system's evaluation process, suggesting significant advancements in AI safety measures.
Read full post on simonwillison.net →
AUTHOR