Hidden instructions in uploaded documents can steer LLM judgements, even when detection fires

Hidden instructions in documents steer LLM judgements

A peer-reviewed study shows hidden instructions in documents can manipulate LLM outputs in CV screening. Detection does not neutralise the effect.

iamvera.ai

A peer-reviewed study shows hidden instructions in documents can manipulate LLM outputs in CV screening. Detection does not neutralise the effect.

Read more: https://iamvera.ai/blog/indirect-prompt-injection-documents-llm-workflows/