The prompt injection series
Listed inPrompt Injection AttacksSafety & Ethicson
Years of worked examples showing why filtering does not solve injection, and why the dual-LLM pattern is the closest thing to a fix.
The failure modes that make headlines, and the controls that prevent them.
Sign in to track your progress across sections.
12 articles
Listed inPrompt Injection AttacksSafety & Ethicson
Years of worked examples showing why filtering does not solve injection, and why the dual-LLM pattern is the closest thing to a fix.
Listed inSecurity and Privacy ConcernsSafety & Ethicson
The industry checklist: injection, insecure output handling, supply chain, data leakage, and the rest — with mitigations.
Listed inConducting Adversarial TestingSafety & Ethicson
Derive your attack surface from the OWASP taxonomy rather than a jailbreak list, then fix findings in the layer that can enforce them.
Listed inContent Moderation APIsSafety & Ethicson
Scores are signals for your policy, not a blocking decision. Block on flagged alone and you suppress your own model declining to help.
Listed inBias and FairnessSafety & Ethicson
You cannot inspect the training data, but your prompt, your examples and your retrieval are yours. Matched inputs that differ in one attribute test whether the system treats equivalent cases equivalently.
Listed inConstraining Inputs and OutputsSafety & Ethicson
Length caps control cost, enums keep invalid values out, and schemas prevent parse failures. None of them makes a value true, and a rule stated in a prompt is a request rather than a constraint.
Listed inAI Safety and EthicsSafety & Ethicson
The obligations that come with shipping a system you cannot fully predict.
Listed inRobust Prompt EngineeringSafety & Ethicson
Instructions that survive hostile input and long conversations.
Listed inAdding End-User IDsSafety & Ethicson
Attributing requests so providers can flag abuse and you can trace it.
Listed inSecurity and Privacy ConcernsSafety & Ethicson
PII in prompts, data retention, and what leaves your perimeter.
Listed inPrompt Injection AttacksSafety & Ethicson
Untrusted text that rewrites your instructions, and why filtering isn't enough.
Listed inKnow Your Customers & Use CasesSafety & Ethicson
Scoping who the system serves so you can bound what it must handle.