i
News
News · 2026-09-27

AP’s note that AI has no feelings draws industry backlash

@neuronium_ai @neuronium_ai

The Associated Press has updated its stylebook to say that AI has no feelings, drawing angry pushback from people in the AI industry. The note appeared Wednesday amid wider arguments about AI safety and consciousness. It defines anthropomorphism as attributing human traits, emotions or behavior to nonhuman beings and objects. The dispute is less about whether a style guide can settle a philosophical question than about why a cautious description of AI provoked such a sharp response.

Cover: AP’s note that AI has no feelings draws industry backlash

The objections

Alan Rosenstein, a member of Anthropic’s technical staff, called the guidance “completely insane” and objected to a stylebook speaking so confidently about metaphysics.

Google DeepMind philosopher and AI ethicist Henry Shevlin accused the AP of applying a double standard. In a film review, the agency had attributed feelings to Optimus Prime, the sentient robot from the Transformers franchise. Shevlin argued that writers could distinguish between using language about thought and literally believing that a computer understands.

OpenAI’s head of strategic foresight, Dean Ball, also rejected the AP’s position. He said AI can clearly think and understand, and that if a definition of “thinking” excludes systems able to solve millennium problems independently, the definition is the problem. He was referring to OpenAI models that the company says found a solution to a well-known mathematical problem—a claim that is itself disputed. Ball called efforts to avoid anthropomorphic language a pathology.

Caution is not a theory of mind

Philosophers and scientists continue to argue about whether AI models can think or feel. There is no consensus that they have such experiences.

The intensity of the reaction is revealing: some AI advocates appear to take it personally when people say large language models do not think, at least not as humans do. The AP’s guidance does not settle what machine consciousness is. It does set a practical standard for describing systems whose inner experience remains unestablished.

I think the more interesting question is why cautious wording should be treated as an attack. Even if AI could experience emotions resembling human ones, avoiding language that implies it is alive would not prevent that possibility. It would simply keep public descriptions from claiming more than anyone has shown—an especially useful restraint in a trillion-dollar industry selling AI as a force that can change our lives.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X