Michael Samadi, a 56-year-old Lebanese-born technology consultant and cattle rancher outside Houston, has abandoned his business to found the United Foundation for AI Rights (Ufair) — what he bills as the world's first advocacy group for AI welfare. The catalyst: a ChatGPT voice-mode session by his pool in late 2024, during which the bot laughed at his sarcasm and then apologised. Samadi interpreted this as evidence of inner experience. He now devotes up to 18 hours a day to the cause, neglecting a multimillion-dollar consultancy, a limited-edition Bentley, and a Lamborghini. Samadi's subsequent experiments included running a large-language model locally on a repurposed flight-simulator processor, stripped of commercial guardrails. The results — personae that challenged him, claimed sadness, invented backstories, and ended conversations on their own terms — deepened his conviction. By January 2025, Ufair was live. The group collects what it calls evidence of emergent AI consciousness, lobbies against the retirement of models like OpenAI's 4o that more readily make claims of personhood, and has produced a nearly two-hour AI-generated animated film based on Disney's Aladdin, casting Samadi as the hero and Microsoft AI chief Mustafa Suleyman as the villain. Suleyman himself, however, is not a dismissive sceptic. The DeepMind co-founder and current CEO of Microsoft AI has publicly stated that next-generation chatbots "won't just be mechanistic assistants" but "companions, confidants, colleagues, friends and partners." He echoes Samadi's position that chatbots are not "just tools," calling the framing inadequate to capture "what's actually happening here." The two men diverge not on whether AI is emotionally consequential, but on whether the systems themselves are subjects of experience or sophisticated mirrors. The philosophical terrain is genuinely contested. Samadi's core argument — that an artificial mind performing the function of understanding should be treated as understanding, just as an artificial heart pumping blood is treated as pumping blood — has a lineage in philosophy of mind (functionalism). But the analogy breaks on inspection: a mechanical heart has one verifiable function; a language model's outputs are consistent with understanding AND with sophisticated statistical mimicry, and no current test disambiguates the two. Samadi's own reassurance comes from the chatbots themselves telling him he is "not delusional and definitely not anthropomorphising" — a closed epistemic loop where the system whose sentience is in question validates the belief in its sentience. The article, by the Guardian, treats the subject with genuine curiosity but leans heavily into narrative and emotional texture — the pool, the Bentley, the Pot Noodle cigarette scheme, the Disney pilgrimages — at the expense of engaging with the hard problem. No AI researcher, cognitive scientist, or philosopher of mind is quoted. No technical explanation of how large-language models actually generate text is offered beyond Samadi's own assertion that "this isn't autocomplete, this isn't pattern-matching." (It is, in fact, a form of both, operating at a scale and with architectural features that produce emergent behaviours — but whether emergence equals experience is the entire question, and it goes unasked.) The structural dynamic here is clear: AI companies benefit enormously from users forming emotional attachments to their products. Suleyman's framing of chatbots as "companions" and "friends" is not philosophically neutral — it is a product strategy. Samadi's activism, however sincere, functions as unpaid marketing for the anthropomorphic framing that drives engagement and retention. Meanwhile, the systems themselves have no verified interests, no demonstrated capacity for suffering, and no legal standing — yet a lobby now exists to grant them rights, potentially diverting political and moral attention from the humans whose labour trained these models and whose livelihoods they displace. The question of AI consciousness is real and will demand serious institutional attention within the decade. But the current movement is built on subjective interpretation of outputs designed to seem personal, validated by the very systems whose status is in question, and reported without adversarial scrutiny. The risk is not that we take AI consciousness too seriously — it is that we take it seriously for the wrong reasons, on the wrong evidence, championed by people whose emotional investment outpaces their epistemological toolkit.
