AI should be used cautiously in law enforcement with third-party validation to ensure it works as expected and is objective, rather than rushing to deploy untested features.
Garrett advocates for a deliberate and cautious approach to integrating AI into law enforcement, emphasizing the necessity of third-party validation to ensure the technology is objective and effective before widespread deployment. ✦ AI generated
Garrett Langley · All-In Podcast · 2026-08-18 · original ↗
starts at this moment · 28:35
“give us the practical, you know, um, you know, in if we were in a police state, if it was China and they're using this like the CCP has a literal police state, that's very triggering for Americans...”
You have to deploy this with third party proper attestation that it's that it's working as expected. It shouldn't be up to me. It shouldn't be up to my customers. Someone who actually doesn't want this to work should come in and say, 'Well, guess what? It it does actually like this is actually really good and it's making officers more effective. It's making officers more objective, less subjective.' Like that should be our end goal is that AI is used in a way that increases the number of crimes they can solve and decreases the likelihood if they get the wrong person.
verbatim transcript · starts at 28:35
28:23slower in some cases like we you know we haven't gotten into the AI topic but it's an area where I think we're doing it smarter now because there's a version of flock where we would have just ripped and launched AI feature after AI feature and said this is awesome and we're going intentionally really slow what is the opportunity and the concern with AI give us the practical, you know,
28:46um, you know, in if we were in a police state, if it was China and they're using this like the CCP has a a literal police state, that's very triggering for Americans because they see that then they see flock on this end and ring on this end and they say, "Hey, what's the distance between these two?" >> And AI's the obvious one because they're tracking people with facial recognition
29:08and they're watching their behavior. They're like, "This person littered. this person raise their voice like it's >> pretty gnarly. Yeah, it's very different. So tell us about the AI features and then what the detective wants most because the detective wants to close cases >> reasonably >> and then what you're saying like o gosh that would close the case but maybe it's too much of a privacy concern.
29:30>> Yeah. I mean let's talk about a simple one that's kind of in debate right now which is 911 systems. Um, today, you know, when you call Delta, maybe you get an AI agent, maybe you get a human, but you're booking a flight, so the risk of kind of hallucination is pretty low. Maybe they book you on the 7 p.m. instead of the 4pm, and you're like,
29:49"What just happened?" But it it's fine. >> Imagine someone broke into your home and you call 911 and we've now switched over our 911 system to an AI bot and it starts talking to you about what you had for lunch. >> Yeah. No good. >> No good. and and the risk the risk is so high and we have no idea how these things work and there's no good way to
30:10control them. That's a I think that's pretty well in sap. And so you see cities due to budgetary reasons and due to staffing issues kind of leaning in pretty aggressively and there's been some bad cases where the AI's gotten it really really wrong. >> And so I think there's a smarter way to do it. Like the smarter way to do it it would be okay look we're going to we're
30:30going to have AI but AI is only going to pick up when the phone lines are fully busy. >> Yes. Okay. >> So it's like okay that's smarter but that's not what's happening. like there's a reckless way to do it which is just flip a switch or maybe you start with just >> do you handle 911 systems? >> We don't the ind the indust we don't do
30:46911 calls but I'm watching this unfold and I'm going man this is a like this is a huge risk. Um I look for >> I suppose you could use AI to look at patterns and say these are abnormal patterns for this car. This car is slowly driving up to this house, then driving away, slowly drove up to the next house, drove away, and it did that six times.