Executive brief
The mem0 server, which provides a memory layer for AI agents, contains a security flaw that allows anyone on the network to delete stored memory records. Because the system fails to check for user identity or permissions on its deletion interface, an attacker could remotely wipe data from the database. This could lead to significant data loss and disrupt the operations of AI agents relying on that stored information.
Technical details
The mem0 1.0.0 server is vulnerable to missing authentication (CWE-306) and missing authorization (CWE-862) within its memory deletion API. Specifically, the 'DELETE /memories/{memory_id}' endpoint does not verify the identity or permissions of the requester. A remote, unauthenticated attacker can exploit this by sending crafted HTTP DELETE requests to remove any memory entry from the database by its ID. This results in unauthorized data modification and a potential denial of service for AI agents that depend on the affected memory records.
Affected products
- mem0ai mem0 1.0.0
Timeline
- 2026-05-12: disclosed: Initial disclosure of CVE-2026-31244
- 2026-05-12: advisory
- 2026-05-13: other: CISA-ADP enrichment and CVSS scoring added