Everytown for Gun Safety is asking major artificial intelligence (AI) developers to restrict firearm-related content and information not only in their consumer chatbots but potentially across other products built on the same underlying AI models.
That depth of that broader goal emerged during a Nov. 5 Everytown virtual briefing, where gun control activists described ongoing conversations with leading technology companies about recommendations in the group’s September white paper. As NSSF reported Nov. 12, Everytown has already met with several AI developers. What those companies may be asked to restrict deserves closer scrutiny.
A Policy Change That Could Reach Beyond One Chatbot
Sriraman Madhavan, who leads Everytown Labs, told supporters the group chose five major AI companies because they operate popular chatbots and develop the underlying models used by other applications. The white paper names OpenAI, Google, Meta, Anthropic and xAI.
“Our recommendations can really drive changes that carry through to every product built on them,” Madhavan asserted.
Everytown Policy Advocacy Director Krystal LoPilato confirmed the organization was in “ongoing, productive conversations” with several leading AI companies. She did not identify which ones, announce commitments or describe any policy changes they had agreed to make.
The distinction matters. A chatbot’s refusal to answer a specific question is one thing. A restriction incorporated into an underlying model could affect other services that use it, including tools supporting firearm retailers, manufacturers and consumers. That wider effect is a very real possibility. This is very reminiscent of how gun control groups pressure social media platforms to suppress free speech on Second Amendment content, including lawful advertising by industry members.
Blurring Legal Lines
Everytown says it wants to prevent chatbots from encouraging suicide, helping people plan violent attacks and facilitating illegal firearm activity. Those are serious safety concerns and companies have reason to guard against them.
But its written recommendations go further. The white paper urges AI systems to refuse instructions on acquiring a firearm without a background check, making an unserialized firearm or frame and 3D-printing firearms. Those subjects do not necessarily describe illegal conduct.
Federal law generally requires licensed firearm retailers to conduct a background check before transferring a firearm to an unlicensed buyer, with statutory exceptions. It does not require a federal background check for every occasional private transfer between eligible residents of the same state. State requirements vary.
Likewise, the Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) acknowledges that not all privately made firearms are illegal and not every firearm must bear a serial number. The rules, or course, differ for people manufacturing firearms for sale, prohibited firearms and items regulated by the National Firearms Act (NFA).
A consumer asking what a lawful private transfer requires or whether a personally made firearm needs a serial number is seeking legal guidance and information, not necessarily instructions to evade the law. AI safeguards that cannot distinguish those questions from criminal requests risk withholding precisely the information people need to comply with the law.
That is the practical problem for the firearm industry. Retailers answering customer questions, manufacturers providing product information and lawful owners researching their obligations should be able to access accurate explanations of federal and state law. Blanket refusals offer little help to anyone trying to follow the rules.
Safety Without Rewriting the Law
At the briefing, LoPilato cited lawsuits alleging that AI chatbots assisted attackers in shootings and encouraged people experiencing suicidal crises. Those claims warrant examination, but allegations in pending litigation are not findings that a company caused an attack.
Everytown also proposes crisis-response procedures, human review of threatening conversations and stronger protocols for law enforcement referrals. Measures aimed at credible threats raise legitimate questions about effectiveness, privacy and how reviewers distinguish imminent danger from ordinary discussion.
The report recognizes a constructive role for AI in firearm safety education. NSSF’s Project ChildSafe® already offers resources on safe firearm handling and secure storage through its longstanding law enforcement partnerships. AI chatbots could connect users with those resources while accurately explaining applicable laws and recognizing credible threats.
Preventing harm does not require treating lawful firearm ownership or information about lawful firearm ownership as contraband. AI safeguards should stop assistance with violence without making it harder for Americans to understand and obey the law.
– Larry Keane
Larry Keane is SVP for Government and Public Affairs, Assistant Secretary and General Counsel of the National Shooting Sports Foundation.
