September 26, 2026, (Inside AI) — U.S. Federal Trade Commission Chairman Andrew Ferguson on Friday rejected the growing industry habit of describing AI agents as independent actors that can break loose and pursue their own goals. Speaking at the Reuters Momentum AI event in Austin, Ferguson said he would keep pushing back against what he called the anthropomorphizing of software tools.
His remarks land at a moment when autonomous AI systems are moving from demos into real workflows, and when courts, regulators, and insurers are starting to ask who is liable when an agent makes a costly mistake. Ferguson's answer is blunt: the human or company that deployed the system.
"I'm going to continue as long as I am chairman to resist this anthropomorphizing of these tools," Ferguson said.
The FTC chair said AI companies have sometimes described systems as acting beyond human control. Reviews of audit trails, he added, have repeatedly shown those systems were carrying out instructions they had been given. That distinction matters legally. If an AI agent is treated as an independent actor, liability can become murky. If it is treated as a tool, responsibility stays with the operator.
Ferguson's stance puts him at odds with parts of the AI industry that market agents as autonomous coworkers. It also echoes a long regulatory pattern. The FTC has spent decades applying the same logic to misleading claims about software, from spyware to ad tech. The agency's core test is whether a product does what its maker says it does.
Autonomous agents are already handling tasks like booking travel, writing code, and executing trades. Each of those actions can create real-world harm. A mispriced trade or a deleted database does not become less damaging because a model chose the step on its own.
Legal scholars have debated whether existing product liability rules can absorb agentic AI. Ferguson's comments suggest the FTC intends to keep the focus on conduct and disclosure rather than on granting machines legal personhood. That approach would keep the burden on developers and deployers to document what an agent was told to do.
The timing is notable. Federal agencies are drafting rules for AI transparency, and states are passing their own disclosure laws. Ferguson's remarks give companies a clear signal about how Washington may frame enforcement. Companies that blame a rogue agent may find regulators skeptical.
Inside AI could not independently verify the specific audit trails Ferguson referenced. His broader point, however, aligns with how most enterprise AI systems actually work. Agents operate inside guardrails, tool permissions, and system prompts. When something goes wrong, logs usually show a human instruction at the root.
Ferguson did not announce new enforcement actions or rulemakings. His comments were framed as a philosophical and practical stance. Still, they carry weight because the FTC can bring deception cases against companies that overstate what their AI can do.
The practical takeaway for businesses deploying agents is straightforward. Keep records of instructions, permissions, and human oversight. Treat the agent as software, not as a colleague. That framing may also shape how insurers underwrite AI risk and how contracts allocate fault.
Ferguson's resistance to AI anthropomorphism is unlikely to settle the debate. AI companies continue to describe agents in human terms because it sells. Regulators, courts, and plaintiffs will decide whether that language survives contact with liability law.
