Biometric matching improves personalized memory in shared AI assistants
Personalizing LLM Agent Memory Using Biometrics
Artificial Intelligence
Summary
AI assistants that remember things about users can work better when they know who is asking for information. The authors designed a system called Bio-Memory that uses biometric signals like face or palmprint data to make sure the assistant recalls the right memories for the right person. This method helps the assistant avoid mixing up information between multiple users sharing the same system. They tested Bio-Memory on different datasets and found it could reliably tell user identities and improve personalized memory retrieval.
What this means in practice
- •For smart device developers: Add biometric authentication to memory retrieval in AI assistants shared by multiple users to personalize responses securely.$Commercial implications: Enables AI assistant makers to offer secure personalized experiences in devices used by multiple people, improving privacy and relevance.
- •For security software teams: Integrate biometric-based identity filtering in shared AI applications to prevent unauthorized access to user-specific memory data.
Authors
Yanhong Qian, Qingguo Meng, Shihao Ding, Xingbo Dong, Zhe Jin, Hanrui Wang, Isao Echizen
Abstract
Personalized memory helps LLM agents deliver stable, tailored assistance by storing and reusing user-specific data across interactions. In multi-user scenarios, however, retrieval must consider not only semantic similarity but also whether the current requester matches the identity associated with the stored memory. We propose Bio-Memory, a biometric-aware memory architecture that conditions memory retrieval on both semantic similarity and biometric matching. Built on top of A-Mem, Bio-Memory augments each atomic memory note with a biometric embedding and uses biometric matching to form the retrieval candidate pool before semantic ranking. We evaluate Bio-Memory on LoCoMo in a 10-user shared-agent setting over 7 face benchmarks and 10 palmprint protocols. Across datasets, Bio-Memory consistently separates owner and non-owner queries. Under face-based personalization, the largest average gap reaches 27.29% / 21.15% in F1 / BLEU-1 on CALFW; under palmprint-based personalization, the corresponding gap is 25.75% / 19.22% on MS_Blue. These results support biometrics as a practical control signal for personalized memory retrieval in shared environments.