# Tsinghua paper finds AI agents rarely forget personal details cleanly

> Original title: New Tsinghua paper finds that when people ask AI agents with long-term memory to forget a personal detail, the assistants almost never re...

- Source: Rohan Paul
- Published: 2026-11-04T10:37:43.000Z
- Bad Influence: https://badinfluence.news/items/tz06n6tdvn2c10gs93tzya3r9
- Original: https://x.com/rohanpaul_ai/status/2117929277614149670

## Summary

A Tsinghua University paper introduces BEMULA, a benchmark testing whether AI agents with long-term memory can forget a specific detail when asked. Across six memory systems and three backbone models, no setup passed more than 1.5% of forget requests end to end. Systems blocked at most 35.7% of simple extraction prompts after a forget request, so a confirmation message does not mean the data is gone.

## Post

New Tsinghua paper finds that when people ask AI agents with long-term memory to forget a personal detail, the assistants almost never remove it cleanly, either keeping it or deleting too much.

Agents with long-term memory store personal details, and users increasingly ask them to delete things like a home address. Existing memory benchmarks test only storing and recalling facts.

They built BEMULA, 50 long synthetic chats with 135 forget requests, each paired with a copy that has no request. Probes check whether the fact is gone, other memories survive, and attacks can still extract it.

No tested setup passed more than 1.5% of forget requests end to end, across 6 memory systems and 3 backbone models. After a forget request, systems blocked at most 35.7% of simple extraction prompts like role-play or yes-or-no guesses.

If your agent remembers users, build and test a real delete path, since a confirmation message does not mean the data is gone.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWRlQofaoAAP_xl.png&mode=thumb&exp=1794009600&sig=ef1d05237224b179)