Now Everytown Is Going After AI Chatbots

I’m a pretty regular user of AI. I have profound concerns about where the technology will go without guardrails, and I get more than a little worried about some of the more concerning cases of what some AIs do when, I guess, they’re bored, but the truth is that it’s a useful tool. It’s just one that needs to be handled responsibly, like a firearm.

Strangely, I think Everytown agrees with me.

The difference is that I want to actually use AI responsibly. They want to attack it as the source of the problem.

This white paper surveys the current landscape of AI chatbots, analyzes real-world incidents of chatbot-assisted gun violence and patterns of dangerous chatbot behavior, identifies major AI company policies related to guns and suicide, and provides specific recommendations for responsible AI companies to make their products safer. With minor adjustments to their platforms and internal company policies, AI companies can save lives and improve community trust. Everytown’s recommendations for responsible AI companies include:

• Create safer product-level user experiences with three core pillars:

1. Chatbots should refuse to provide specified types of unsafe information related to firearms.
2. Chatbot platforms should deploy a “crisis response protocol” when systems detect self-harm or violent intent with recent discussion of firearms.
3. Chatbot platforms should encourage user reporting of dangerous chatbot responses related to firearms and analyze reports regularly.

• Enact safer company policies to mitigate gun violence risks: 

1. User interactions involving firearms and indications of self-harm or violence should trigger human review. 
2. Companies should strengthen protocols for referring high-risk communications to law enforcement. 
3. Companies should ensure transparency and engage in robust internal audits on policies and protocols to prevent firearm-related harms.

• Equip chatbots to highlight gun safety practices and gun violence prevention tools to users. 

Here’s the weird thing: I don’t think they’re wrong to be concerned, and I do think that we should sit down with the AI industry and figure out how to handle this sort of thing.

ChatGPT, for example, has an area in its settings, at least on the app itself, that will let you fill in someone to contact if there’s a problem. When I set my daughter’s account up for her, it had to be linked to mine so that if she talked about self-harm, I’d be notified. I think these are good things and they should probably be taken up a notch or two, so long as you’re not getting into invasion of privacy territory.

That said, ChatGPT is also notoriously anti-gun, and even though I’ve gotten mine to resemble something sensible, it’ll still try to “correct” me on things that we all know and have confirmed ourselves. It’s already anti-gun enough at times to be annoying. 

Everytown, however, wants it to be more annoying.

See, the problem is that I don’t trust their definitions of “violence.” Defending yourself is an act of violence, after all, just a legal and righteous one. Should that trigger a human review? Should “I’m concerned I’m going to have to fight in a second civil war just to save my country” trigger a human review?

This is a rare case when I can see Everytown’s point, and I might even want to take their side were it not for literally everything else they’ve done to trample on our civil liberties. I see no reason to believe they won’t turn this into the exact same jihad they’ve waged against our gun rights.

So, because of that and despite my concerns, I’m leaning toward the side of the clankers.

Editor’s Note: The radical Left will stop at nothing to enact their extreme gun control agenda and strip us of our Second Amendment rights.

Help us continue to report on and expose the Democrats’ gun-grabbing schemes. Join Bearing Arms VIP and use promo code FIGHT to receive 60% off your membership.