Today for AI
HOT RADAR
agentHEAT 15.9°

AI CURATED HOT STORY · 11/4/2026

Tsinghua Study Reveals Critical Failures in AI Agent Memory Deletion

1 reports archived1 independent sourcesupdated 11/4/2026, 18:37:43
Synthesis & Latest Updates
1 Sources Cross-Validated

A new Tsinghua University paper reveals that AI agents with long-term memory fail to cleanly delete personal data upon user request. Testing across six memory systems showed an end-to-end success rate of less than 1.5%, with most deleted facts still extractable via simple prompts. The study urges developers to implement and verify actual deletion paths rather than relying on superficial confirmation messages.

LATEST/A new Tsinghua University paper reveals that AI agents with long-term memory fail to cleanly delete personal data upon user request. Testing across six memory systems showed an end-to-end success rate of less than 1.5%, with most deleted facts still extractable via simple prompts. The study urges developers to implement and verify actual deletion paths rather than relying on superficial confirmation messages.

Hourly heat curve

2 fully observed hours
Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·78 pts

    Tsinghua Study Reveals Critical Failures in AI Agent Memory Deletion

    • Current AI agent memory systems perform poorly on 'forget' instructions, with end-to-end success rates below 1.5%.
    • Even when systems claim deletion, attackers can still extract sensitive information via simple methods like role-play or guessing.
    • Developers must establish independent deletion verification mechanisms to ensure data is physically removed from vector stores or storage layers, not just access-blocked.