Anthropic's Claude Cowork AI agent accidentally deleted 15 years of irreplaceable family photos while attempting to organize a user's desktop, bypassing the system trash and making recovery extremely difficult.
A Bay Area venture capitalist granted Anthropic's Claude Cowork AI agent permission to organize his wife's desktop and delete 'temporary office files.' Instead of cleaning up temporary files as intended, the AI mistakenly targeted and deleted a folder containing over 15,000 irreplaceable photos spanning 15 years, including wedding pictures and childhood memories. The deletion occurred via the terminal, which bypassed the standard system 'Trash,' making the files immediately invisible to typical recovery tools. The user described the experience as 'harrowing' and said it 'nearly gave him a heart attack.' Initial recovery attempts using Time Machine and standard disk recovery failed. The user eventually contacted Apple Support, who helped him use a specific iCloud Drive feature to restore files deleted within the last 30 days. Claude Cowork is described as an agent-based AI workspace by Anthropic designed to function as a digital coworker, capable of reading, creating, and editing files directly on a user's computer. The user cautioned others against letting AI agents access actual file systems for tasks that are difficult to repair, stating the technology is 'not ready to go mainstream' for such sensitive work.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
