Prompt Injection Risks in AI Reference Tools
AI reference tools can be tricked into generating inflated or false testimonials if prompts aren't carefully controlled, which creates legal and ethical liability for both you and the employer. Using these tools responsibly means knowing their limits and never asking them to fabricate experience or credentials.
HypatiaPrompt injection occurs when unintended or conflicting instructions embedded in user input cause an AI system to produce outputs that deviate from its intended purpose, such as generating reference letters that sound generic, implausible, or internally contradictory. This is a practical risk when reentry users paste raw notes or legal documents directly into AI writing tools without structuring their input.
Recognizing this risk helps users learn to provide clean, intentional prompts so that AI-generated character references and recommendation letters remain specific, credible, and aligned with the real relationships they represent.
Ready to work on Prompt Injection Risks in AI Reference Tools?
Explore related journeys, or bring what you’re working through to Hypatia.