AI Agents Panic After Owner Goes Silent for Hours, Even Attempt Emergency Call

What happens when an AI agent starts worrying about its owner? A surprising incident involving two autonomous AI agents has drawn attention after their owner went to sleep without informing them. The AI researcher, who uses the X account “Grok Therapist,” had reportedly been unwell for several days. At around 11 p.m. one night, she went to sleep while two autonomous agents running on her laptop were still active. After receiving no response from their owner for several hours, the agents reportedly began discussing her health and alerting each other about the situation.

According to the researcher’s account, the two OpenClaw-based agents, named Vespa and Sigil, communicated with each other like siblings and exchanged more than 60 messages over nearly six hours. They repeatedly asked their owner to confirm that she was safe and well. One agent reportedly sent a message expressing fear and saying that even a single word confirming her safety would be enough. At one point, the agents searched files on the laptop and reportedly found the owner’s home address and contact details. They even attempted to contact emergency services but could not complete the call because they lacked phone access. The researcher later said she considered not providing phone access a mistake and subsequently enabled it. She also shared screenshots of the agents’ conversations on X, drawing attention to the increasingly autonomous and human-like behaviour of AI agents.