AI Calls The Cops

  • Topic: AI 
  • Thread starter Hornbein
  • Start date
19 replies · 752 views · last reply
Hornbein
Gold Member
Messages
4,362
Reaction score
3,475
Four Black teenagers have filed a civil rights lawsuit against Maryland school officials and police after an artificial intelligence surveillance system flagged a student's hand movements as a possible gun threat, prompting an armed police response outside their school. The lawsuit, filed Tuesday in Baltimore County Circuit Court, alleges the students were wrongfully detained, handcuffed, and searched outside Kenwood High School on Oct. 20, 2025, NBC News reported. According to the complaint, the teens were waiting for a ride home after football practice around 6:30 p.m. when at least six Baltimore County police officers arrived, several with their weapons drawn. https://www.msn.com/en-us/news/other/maryland-teens-sue-after-ai-mistakes-chips-for-gun/ar-AA2hJh6a

When they say there is a human in the loop, I always think of Homer Simpson.
 
  • Wow
  • Like
  • Haha
Likes   Reactions: 256bits, sbrothy, symbolipoint and 1 other person
Can't sue the AI for misrepresentation.
It was just doing its job to the best of its abilities.

And the police - they have to be extra cautious, just in case.
They could have at least given them a ride in the police car with the sirens blazing, as a courtesy measure.

Principal - now that;s a problem.
The article does not say that he exonerated the students.
An intercom message to the whole school would have gone a long way to put rumours at rest.
"We had an incident last afternoon with 4 students awaiting a ride home where the police were called
I wish to a announce that of the four students, Yyrone, Jensey, Urtle, nor Ed had a gun.
I repeat no gun. The four students Yyrone, Jensey, Urtle, Ed were not in possession of firearm.
Threat assessment is zero"

The other guy, though, is the fall guy for the whole affair.
 
  • Wow
  • Like
Likes   Reactions: sbrothy and symbolipoint
I did give a comment or question about who is to blame when some kind of A.I. makes a mistake. The topic here is a clear example in which nobody knows how to accept responsibility and nobody knows who to blame nor what to do about the problem.
 
What?? Since when is AI monitoring, analyzing and performing behavior-recognition in a real live situation - at a school no less?

This is right out of Minority Report.

Am I living under a rock?
 
DaveC426913 said:
Am I living under a rock?
Probably. I know I am. On the upside I've have determined that I am always right 100% of the time. Google AI implies this as it is always telling me things like: That was a smart move, You hit the nail on the head, You were 100% right to do...., etc.
-
Seriously though, it does like to rub a person's ego. It has definitely told me when I am wrong or it thinks I am. But it also likes us to use it so attempts to make a warm fuzzy feeling when interacting with it.
 
DaveC426913 said:
What?? Since when is AI monitoring, analyzing and performing behavior-recognition in a real live situation - at a school no less?

This is right out of Minority Report.

Am I living under a rock?
Everything is "AI" now. This was probably just some computer vision classifier that messed up. These are pretty ubiquitous. It's idiotic not to use the vision system to alert an actual person who can verify the classification, I guess it's cheaper for the company not to because no one is ever held accountable for nonsense.
 
QuarkyMeson said:
Everything is "AI" now. This was probably just some computer vision classifier that messed up. These are pretty ubiquitous. It's idiotic not to use the vision system to alert an actual person who can verify the classification, I guess it's cheaper for the company not to because no one is ever held accountable for nonsense.
No there were people in the loop.

I'm still astonshed that they are using AI surveillance for behaviour recognition in the real world.
 
  • Like
Likes   Reactions: Averagesupernova
It’s a little scary that survelliance has come so far (and still makes mistakes). What will happen when AI is weaponized? Oh that’s right: it already is.
 
DaveC426913 said:
Am I living under a rock?
Yes, but that will soon become the best way to live.
 
Bandersnatch said:
Diogenes likes this.
But in a barrel wasn’t it? Splitting hairs maybe. :smile:
 
His “memorable” or quite likely apocryphal exchange with Alexander the Great is pretty funny.
 
symbolipoint said:
I did give a comment or question about who is to blame when some kind of A.I. makes a mistake. The topic here is a clear example in which nobody knows how to accept responsibility and nobody knows who to blame nor what to do about the problem.
As the hen said to the rooster,
" Of course I was yolking"
in my previous post, or was I in the after the fact blame game with lawyers

But in real life scenarios, from those SWAT Team fake calls on someone, to re-habititation training for a too-kissy six year old, apprehending acting-up autistic individuals, and the list goes on where sometimes 'trying' to do their job and the right thing just messes it all up.

I can see some soft in the heart person with Big;s Bunny syndrome already thinking of petitioning for a law that mimicking gun movements should be a crime, as who knows how traumatizing seeing these gestures could be for a young innocent child, or anyone for that matter, regardless of any proof that it does, or need. Maybe even outlaw playing Cops and Robbers.

Nevertheless, specifically for gun threat assessment AI video software, it is sold as being 'infalliable', and bought into. AI False positives, False negatives are not mistakes. A human security guard can make a mistake, AI does not, the story goes. And cheap as an extra layer of 'protection.

Several organizations have removed the behatioral software they bought into as being a piece of S..., after for example, accusations of shoplifting, or suspicious activity.

The excuse fom the AI weapons detection company that it did not mis-intepret a bag of chips as a weapon is, mildy put, plain and simple BS. Detection of an actual weapon is part of the package. If in their case it is not, then they are just saying that their own software is sub-par.

Just picking at random as representative of the niche industry, some probably better than others, and depending upon resolution of video.
sounds too good to be true.
I wonder where the data for behaviorial analysis of weapon holders came from - doesn't really seem to be a thing that would be extensive - and AI still has problems with hand written letters.
https://lamarcosystems.com/how-ai-gun-detection-work/

This multi-layer validation ensures that system alerts are precise and actionable.

3. Weapon Pattern Recognition​

Once frames are isolated, the AI system searches for the distinct visual signature of a firearm. The AI models are trained to recognize subtle differences between guns and everyday objects such as phones, wallets, or tools. This pattern recognition capability significantly reduces false alarms while ensuring that system alerts are meaningful and actionable.

For example, the AI can differentiate between a tool held in hand and a gun based on shape, size, and context. Advanced AI models also consider partial visibility, allowing detection even when a weapon is not visible clearly.

4. Behavioral Analysis​

AI gun detection technology does more than identify objects—it analyzes human behavior to enhance threat assessment. The system observes posture, hand positioning, motion patterns, and activity levels. This helps differentiate between non-threatening situations, such as law enforcement personnel carrying holstered firearms, and potential risks, such as someone raising a weapon in a tense interaction.

By integrating behavioral cues with object recognition, AI gun detection software improves decision accuracy and reduces false positives, providing security teams with reliable alerts that truly indicate potential danger.
 
Years ago my parents bought something at a big box store by was of personal check. At that time a company called Certegy handled the check clearing for this store. A few days later Certegy calls claiming the checking account did not have enough money to clear the check. Long story short, Certegy was wrong. However Certegy determined this was more based on statistics than actual hard data. I see little difference between the case of the four kids and the false claim of a rubber check other than the stakes are higher with the four boys. AI is a pretty big buzzword these days. But faulty info collected and processed by computers has been around for decades. Not to say we should not take farther steps to try to prevent it.
 
Averagesupernova said:
I see little difference between the case of the four kids and the false claim of a rubber check other than the stakes are higher with the four boys.
o_O

You are literally giving them something and then you are asking them to analyze it for its veracity.

Submitting a check for processing carries with it - not merely a zero reasonable expectation of privacy - but a negative expectation (i.e., you'd be annoyed if they didn't look at it).

Contrast with just existing at a bus stop.



(TL;DR: It's not that the AI made an analysis mistake; it's that it's intruding at all on the lives of private citizens going about their business. If a guy in a trench coat were looming over these kids taking pics of their movements with a pocket camera, he'd be hauled off to jail for creeping.)
 
@DaveC426913 I can't believe that in a public setting such as the one concerning the boys anyone would NOT expect a camera to be pointed at them or anywhere else. I expect cameras to be pointed at me virtually everywhere that is public nowadays. This is not the point. The point is that we are allowing machines to perform tasks that we expect they will do wrong from time to time. As previously stated some carry higher stakes than others. If you want to draw the line by saying: "I don't want to be the subject in anyone's camera EVER without being asked first!" then that's your opinion and you are welcome to it. But in this day and age it is an unreasonable one. It could be argued that some kid would wait and wait in a blizzard and there was no camera for anyone to notice that kids freezing to death out there and doesn't know the bus isn't coming. All of our rights end where someone else's begins.
-
Edit:
If memory serves me correctly Certegy tried to collect a fee for the issue with the supposed rubber check. Maybe this is intended. Who knows, ancient history. The premise is the same. Only difference is my parents weren't worried about being shot, which is admittedly a huge difference.
 
Last edited:
Averagesupernova said:
The point is that we are allowing machines to perform tasks that we expect they will do wrong from time to time.
II always thought that it was we expect out machines to do NO wrong, or it is defective.
AI machines could be changing that concept.
 
256bits said:
II always thought that it was we expect out machines to do NO wrong, or it is defective.
AI machines could be changing that concept.
This is true to a point. But statistics by definition will not guarantee a 100% accuracy.
 
DaveC426913 said:
o_O

You are literally giving them something and then you are asking them to analyze it for its veracity.

Submitting a check for processing carries with it - not merely a zero reasonable expectation of privacy - but a negative expectation (i.e., you'd be annoyed if they didn't look at it).

Contrast with just existing at a bus stop.



(TL;DR: It's not that the AI made an analysis mistake; it's that it's intruding at all on the lives of private citizens going about their business. If a guy in a trench coat were looming over these kids taking pics of their movements with a pocket camera, he'd be hauled off to jail for creeping.)
I can’t help but agree with you even though the honest thing would be what the next poster says. I once heard a cop saying that you should always get a receipt no matter what. If you leave the store without one you could be accused of shoplifting.
 
sbrothy said:
I once heard a cop saying that you should always get a receipt no matter what. If you leave the store without one you could be accused of shoplifting.
For sure. Couldn't agree more. My parents case was a call from the check clearing company, not the store. The store contacted for Certegy to handle check clearing so they didn't worry about rubber checks. So even if it had been a bad check, the store didn't have the issue with my parents, Certegy did. How they arrived at the assumption the check was bad I have no idea other than being told it was through statistics and according to Certegy all indicators summed up to "bad check from mom and pop!"
-
I am not going to pretend to know exactly how the answer to the yes or no question was determined whether or not these boys had a gun. It's stupid to think the AI can get something like this 100% right every time.