Autonomous agents — such as AutoGPT, BabyAGI, and Voyager — plan multi-step goals, use tools, and adapt to new information independently. They represent the frontier of AI autonomy but demand strong governance, revocation, and identity layers to prevent unintended behavior.