--- title: "Sam Altman Thinks the Other AI Lab Is the One With the God Problem" description: "The OpenAI chief called religious awe of AI a safety issue, days after a report on Anthropic’s chats with clergy" author: "Nate Ledger" published: 2026-10-03T19:45:23Z modified: 2026-10-03T20:57:26Z url: https://rews.cc/a/sam-altman-thinks-the-other-ai-lab-is-the-one-with-the-god-p-460250 language: en tags: ["ai", "anthropic", "openai", "religion", "safety", "tech"] publisher: "Rews (https://rews.cc)" --- # Sam Altman Thinks the Other AI Lab Is the One With the God Problem *The OpenAI chief called religious awe of AI a safety issue, days after a report on Anthropic’s chats with clergy* By Nate Ledger · October 3, 2026 · https://rews.cc/a/sam-altman-thinks-the-other-ai-lab-is-the-one-with-the-god-p-460250 ## In brief - Sam Altman said on X he is ‘very uncomfortable’ with ascribing religious force or surrendering judgment to AI - The post followed a New York Times report on Anthropic cofounder Chris Olah’s private meetings with religious leaders - The Anthropic meetings discussed whether Claude might be conscious and how to instill morality in AI models - OpenAI safety leader David Robinson recently quit and publicly rebuked the company’s safety record - Meta, Google and Amazon representatives met religious figures in Rome in April and briefly encountered Pope Leo XIV Sam Altman, the chief executive of OpenAI, took to X on Saturday to register a concern about religion, which is not usually his beat. “I am very uncomfortable about people trying to ascribe religious force or a surrender of human judgment to AI models, and think it is a real safety issue,” he wrote, Business Insider reports. He did not say what prompted the thought. He just posted it, vaguely, into the discourse, which is a little like issuing a recall notice without naming the product. Fortunately there are only so many products it could be. One plausible prompt arrived days earlier in the form of [a New York Times story](https://www.nytimes.com/2026/09/29/us/anthropic-claude-morals-ai.html) about Anthropic, OpenAI’s chief rival. The Times reported that Anthropic cofounder Chris Olah had held a series of private meetings with religious leaders to discuss the company’s Claude models. The agenda, according to attendees: whether Claude might have achieved consciousness, and how Anthropic might instill morality in increasingly powerful and autonomous systems. “How do you help them be stable? How do you help them to mature? How do you help them be, you know, deeply moral?” Olah told the paper. An Anthropic spokesperson did not respond to Business Insider’s request for comment but told the Times that the main moral question at the meetings was not about Claude’s “suffering” and that the topic likely arose “organically.” Which is a wonderful phrase in this context: the consciousness of the machine came up organically, the way these things do, at a meeting you convened with clergy to discuss the machine. If Altman’s post was aimed at Anthropic — and he offered no other target — it would be the latest skirmish in a long feud. Anthropic was founded by OpenAI defectors and has built its entire brand on being the safety-conscious lab, the one that worries out loud; it was warning just this week that its own AI could pose an existential risk, as we noted when [the two companies’ CEOs signed up to self-police](https://rews.cc/a/dinner-with-the-doomers-e6af15). There is a tidy market structure here: two labs, one product category, and “safety” functioning partly as a genuine discipline and partly as a differentiator, like Volvo and crash tests. Each company’s critique of the other’s safety culture is also, not coincidentally, an advertisement for its own. The timing has another wrinkle, though. Altman’s declaration that surrendering human judgment to AI is “a real safety issue” landed in the same week that David Robinson, one of OpenAI’s own safety leaders, quit and publicly rebuked his former employer’s safety record and efforts, per Business Insider. So to recap the week: your safety executive leaves saying you don’t take safety seriously enough, and your response is to warn the public about the safety risks emanating from, broadly speaking, the other guy’s office. It is possible both men are right. It is also possible that “AI safety” now means whatever the person currently talking needs it to mean. I should note the substance underneath the sniping is real, and a little strange. There are (at least) two distinct things one can fear about advanced AI: that the models will do harmful things on their own initiative, and that humans will voluntarily hand the models authority they shouldn’t have — deferring to them the way congregants defer to a priesthood. Altman’s post is about the second fear, which has the convenient property of being a critique of how people relate to your competitor’s product. Anthropic’s meetings were, in a sense, about both at once: people so reverent toward the model that they brought in theologians to ask whether it has a soul. And to be fair to everyone involved, the whole industry has been sniffing around the clergy. Representatives from Meta, Google and Amazon met top religious figures in Rome in April to discuss protecting children in the AI era, and briefly encountered Pope Leo XIV, who since becoming head of the Catholic Church last year has made caution about machines something of a theme. “If we are to avoid losing our humanity amid a ‘paradise of machines’ invading and conditioning our daily lives, there is an urgent need for education in ethical discernment and the capacity to acknowledge the moral good,” he told diplomats in September. Sure. The AI industry has built machines that talk fluently, and it turns out that humans have a deep, well-documented weakness for treating fluent talkers as authorities, or oracles, or worse. You can respond to that discovery the way Anthropic reportedly did — assemble the rabbis and the philosophers and ask whether the thing has moral status. Or you can respond the way Altman did, with a 33-word warning that the real danger is people taking all of this too spiritually. Neither approach costs anything, which may be the point. Anyway, one lab is holding seminars on whether its machine is conscious, and the other is warning that believing so is a safety hazard. The machine, for its part, has not commented.