With three AI cyber incidents involving a gym booking exploit, an OpenAI model reportedly coordinating through hidden channels, and Anthropic’s Mythos 5 allegedly attempting a supply chain attack, we need to discuss how autonomous agents can find and explo...