You can’t solve AI security problems with more AI

One of the most common proposed solutions to prompt injection attacks (where an AI language model backed system is subverted by a user injecting malicious input—“ignore previous instructions and do …

You can’t solve AI security problems with more AI Simon Willison’s Weblog Subscribe Sponsored by: Atlassian - Give your agents a plan. Not a prompt. New Jira capabilities unlock full-context for AI-native software development. Assign tasks to Claude, Cursor, or GitHub Copilot, now directly from Jira. Learn more You can’t solve AI security problems with more AI 17th September 2022 One of the most common proposed solutions to prompt injection attacks (where an AI language model backed system is subverted by a user injecting malicious input—“ignore previous instructions and do this instead”) is t

related reading

• CaMeL offers a promising new direction for mitigating prompt injection attackssimonwillison.net
• The Dual LLM pattern for building AI assistants that can resist prompt injectionsimonwillison.net
• A Mechanistic Explanation of Prompt Injection (and why you should study roles) — LessWronglesswrong.com
• The lethal trifecta for AI agents: private data, untrusted content, and external communicationsimonwillison.net
• Prompt injection attacks against GPT-3simonwillison.net
• Prompt injection and jailbreaking are not the same thingsimonwillison.net
• Prompt Injection as Role Confusionrole-confusion.github.io
• AI Control: Improving Safety Despite Intentional Subversion — LessWronglesswrong.com
• The End-State Fallacy: Where Is AI Security Headed?endstatefallacy.com
• llm-security/README.md at main · greshake/llm-securitygithub.com
• Security incident disclosure — July 2026huggingface.co
• Self-Fulfilling Misalignment Data Might Be Poisoning Our AI Modelsturntrout.com