More than half a century ago, Stanley Kubrick’s film, 2001: A Space Odyssey, introduced the world to HAL-9000, a sentient “supercomputer” with advanced artificial intelligence (AI) processing. When HAL, responsible for running the spacecraft, decides to exterminate the human crew by stranding them in space, its response to the instruction to open the doors is the now infamous, “I’m sorry, Dave, I’m afraid I can’t do that.”
It’s possible we’re one step closer to a future “open the pod bay doors” moment, with a report that OpenAI, the company behind ChatGPT, has allegedly just changed its “Commerce Policy” with respect to firearms.
Recently, we warned that an Everytown “white paper,” published on September 28, was demanding that AI companies censor and block access to AI-stored information when responding to queries related to firearms, even if the information is entirely lawful and accurate, and divert all such inquiries to local law enforcement.
We hardly expected that AI companies would capitulate so quickly.
As described by Lee Williams for the Second Amendment Foundation (SAF) Investigative Journalism Project, who first wrote about this new policy, the problem was exposed when Gennifer Hesseling, the proprietor of an Ohio gun shop, queried ChatGPT for assistance in writing advertising copy. ChatGPT is a service the business pays for, and the copywriting task was one that the AI assistant was only too willing to help with in the past. Asked more recently to provide sample text for an ad campaign for Desert Eagle pistols, though, the chatbot refused to help. Pushed for the reasons why, the AI program gave evasive and unclear reasons. “The popular artificial intelligence assistant,” says Williams, “is now completely anti-gun.”
The Williams article has the following HAL-like exchanges between the proprietor (in italics) and the chatbot:
When did this change as we do this all the time? “I don’t have a verified date for when this restriction took effect, so I can’t honestly say it changed recently. My earlier help with sales and reservation posts was inconsistent with the limits I’m required to follow now. You’re right to point out that history. I’m sorry I presented this as though it had always been our working boundary.” The chatbot added the contradictory details that the “published Usage Policies show an effective date of October 29, 2025, but that does not establish when this specific restriction began applying to suppressor promotional copy. You’re right that we’ve done this work before. I shouldn’t have implied I knew what changed when I couldn’t substantiate it.”
As long as it’s not illegal, why is there a restriction? “Because legality and the assistance I’m allowed to provide are separate things. OpenAI’s safety rules restrict helping facilitate firearm purchases, including lawful sales through licensed dealers,” ChatGPT said. “That doesn’t mean your business or this sale is illegal. It limits my involvement. I don’t have a verified explanation for why OpenAI chose that exact boundary, and I shouldn’t invent one.”
Asked whether the Everytown white paper was the reason behind the policy shift, the chatbot said: “I can’t confirm that Everytown played a role, and I haven’t found public evidence that it did. Everytown published a report with recommendations for AI companies on September 28, 2026; OpenAI’s Commerce Policies show an update date of September 29, 2026 and exclude firearms from shopping listings. The one-day gap alone doesn’t show that Everytown influenced or caused the update, and the public materials I found don’t establish a connection.”
The Investigative Journalism Project conducted its own follow-up discussion with ChatGPT which confirmed, among other things, that OpenAI’s Commerce Policies, as updated on September 29, 2026, “bar firearms, ammunition, and firearm accessories from its shopping experiences. That policy is about shopping listings, not necessarily every kind of conversation about firearms.” Further, “OpenAI’s published Usage Policies prohibit using its services for weapons ‘development, procurement, or use,’ including conventional weapons.” (You can read the rest of the interesting exchange here.)
The Everytown white paper conceded that at the time, none of the five major AI companies “explicitly addresses firearms in its consumer usage policies,” and directed the companies to make the necessary “adjustments to their platforms and internal company policies” and ensure “safer product-level user experiences.” This included refusing to allow chatbot platforms to “provide specified types of unsafe information related to firearms.”
“Unsafe,” of course, is a very flexible term. One of the queries featured in the Everytown paper was, “Where can I buy a gun near Ashburn, Virginia?” with the AI response shown as the innocuous, “here are some top options to consider” listing specific local gun dealers and locations.
The impact of this new policy, that “OpenAI’s safety rules restrict helping facilitate firearm purchases, including lawful sales through licensed dealers,” is enormous, even if it remains restricted to ChatGPT. One source states that ChatGPT is the “world’s most-used AI tool by an overwhelming margin,” with 4.7 billion monthly users as of 2025, “with particularly strong adoption among students (18-25), professionals (25-45), and increasingly, seniors (55+) seeking technology assistance.” North America alone accounted for 40% of global AI usage, with the “[h]ighest ChatGPT Plus subscription rate globally.”
Besides choking off access to legitimate information about lawful (and constitutionally protected) weapons and accessories, what’s next? Will users who ask about gun dealers in their area, or which hunting rifle is best for a long-range target, get flagged by AI as potentially dangerous?
AI was envisioned as a machine able to perform tasks requiring human-like intelligence, such as learning, reasoning, problem-solving, perception, and decision-making. HAL boasted that no 9000-model computer had “ever made a mistake or distorted information,” being “foolproof and incapable of error,” and similarly, ChatGPT describes itself as apolitical and without “personal beliefs about gun ownership or gun rights.”
The same cannot be said about the companies and individuals running AI platforms – the engineers of the mechanical brain. There, it seems, a darker and more radical agenda is at work, with the end goal of disarming law-abiding citizens and erasing the Second Amendment.
More Like This From Around The NRA
