Artificial intelligence data privacy concerns what information a model or agent can access, where that information is sent, how long it is retained and who can retrieve it. Prompts, screens, files, tool results and generated outputs can all contain sensitive material.
Privacy controls include data minimization, purpose limits, access restrictions, secure storage, local processing and transparent retention rules. Increasing an agent's device access can improve usefulness while also increasing the amount of information that must be protected and governed.
Acronyms and aliases
AI data privacy acronymAI privacy acronymartificial intelligence data privacy variant
Related terms
Frequently asked questions
Why can AI agents create new privacy risks?
Agents may combine screen, file, account and tool access across tasks, exposing more information than a narrowly scoped application or single prompt.
How can AI data privacy be improved?
Use the minimum necessary data, restrict permissions, prefer controlled processing, limit retention, protect logs and give users clear information and choices.
