Bad Memories Still Haunt AI Agents

Cisco discovered and resolved a major vulnerability in Anthropic's memory-handling system. Experts caution that improper management of memory files will persist as a threat to AI systems.