#mitigation

#mitigation

Prompt injection exploits LLMs' inability to separate data from instructions, making these attacks hard to fully mitigate and better viewed as a confusable-deputy exploitation.

Prompt injections remain largely unpreventable, forcing LLM providers to rely on reactive, channel-blocking mitigations that require explicit user consent to prevent data exfiltration.