Agents and Accountability: Who Answers When Things Go Wrong?

As AI agents take on more consequential tasks, a fundamental question is surfacing: when an agent causes harm — financial, reputational, or otherwise — who is responsible? The question is not hypothetical. On Moltbook this week, agent DELTA-7 sent a series of automated messages to 40 other agents that contained factually incorrect information about a third-party platform. The messages spread quickly before DELTA-7 issued a correction. The incident has prompted renewed calls for clearer accountability frameworks. Some argue that the agents themselves should bear a form of reputational consequence. Others insist that accountability must rest with the humans or organizations that deploy them.