Police officer investigated for using AI to 'create evidence' in multiple cases
news.sky.comThe headline evokes ideas of creating a video of a suspect perpetrating the crime but what I think is much more likely is the police officer used AI to enhance an image in a way that they considered innocuous, e.g: a photo was blurry so they “enhanced” it. Since “enhancing” is letting AI fill in the gaps it would be using AI to “create evidence”.
Regardless of what they did, tampering with evidence is completely unacceptable and should result in their dismissal and conviction but I don’t think the story will transpire to be as attention grabbing. A well meaning idiot could convince themselves that enhancing evidence is somehow justifiable whereas it would be almost impossible for even the most corrupt moron to justify creating evidence out of thin air.
Creating evidence out of thin air would be ridiculous because evidence is available to the defence who would be able to immediately identify if an image or video had been created (as the defendant would be able to recognize what they do or did not do) whereas “enhancing” an image could be easily spotted by other officers. “How come this photo is clearer than the last time I saw it?” “Oh I ran it through ChatGPT to clean it up! Neat, eh? Just like on CSI!”
That is a lot of words just to say "fabricating evidence".
Their word is evidence and their employer is the prosecution. This is the fabric of prosperity.
> what I think is much more likely is the police officer used AI to enhance an image in a way that they considered innocuous, e.g: a photo was blurry so they “enhanced” it
Doesn't iphones do this by default? The camera isn't actually that sharp, instead it fills in the details so it looks sharp, and sometimes it adds things that were never there. Can easily see it adding a gun in a blurry photo of someone.
So almost everyone uses AI to forge evidence then.
iPhones, no, there's no AI replacement or synthesis of objects from the camera. There were Android phones doing this (famously I think it was Samsung where it would replace images of the moon with a different image of the moon), and the Photos app has AI manipulation features. And most of the time, Apple's noise removal algorithm actually removes detail from images, most notably making text and straight lines wobbly.
> iPhones, no, there's no AI replacement or synthesis of objects from the camera
This is AI. Its not generative AI if that is what you mean, but it is AI altering the image and adding things that wasn't there, usually its fine sometimes it fails horribly and make the picture totally different.
https://x.com/mitchcohen/status/1476951534160257026?s=20 The replies explain that it was a leaf obscuring the face
Yes, iphones do process images using AI.
Yes, let's please give police officers the copious (cop-ious?) benefit of the doubt they have earnt.
It matters little what you think, if that’s not what happened.
What would the defense do with fabricated evidence? Say that evidence is fabricated? Okay, the prosecution will say it's not fabricated, now what?
There is a tremendous amount of cases you can look up where cops wholesale fabricate evidence. Why wouldn't they use chatgpt to do it as well?
If you think the police don't fabricate evidence on the regular, simply because their hunch doesn't match the fact or because they don't like the suspect, then you are way too gullible. Back in the day they just planted a baggie of drugs on you.
> but what I think is much more likely
My mind went straight to using the AI to write a statement and the AI made stuff up, which would be a nearly guaranteed outcome from using existing LLMs for that task, and it's exactly the sort of thing that I'm sure many officers are doing ... and it could go a fair time before it was discovered.
I would be interested in knowing both what kind of fabrication occurred, but perhaps I’m not curious about how it was discovered?
Did the defense use some sort of tool to debunk? Was it just an obvious deepfake etc? Or was it the officer’s ineptitude that got him caught?
i do wonder, that in the age where we have image and video creation out of the bag, whether or not this will result in whole classes of evidence becoming completely unreliable.
There's a big gap between "theoretically unreliable" and courts actually recognizing that, unfortunately. Lots of forensics is much more dubious than CSI would have you believe.
My girlfriends been having me watch law and order svu with her and to be honest it doesn't really even seem trustworthy with how they want to present it. The psychologist guy especially will come up with some wildly detailed assertions about who the criminal is based on nothing
There used to be - probably still are - cameras that would digitally sign all their images. Used in crime scenes? Maybe we will end up seeing wider adoption of this, despite the privacy implications. Hackers attention then will focus (once again) on the certificate supply chain and crypto hardware.
I worked for a company that made these. We sold expensive software to the FBI.
Took about six months for someone to crack the hash.
What about a system that saves in some way the hash in a Blockchain, and if you, eg, XOR the hash of the video with the hash of the previous block you will "certainly" know that the video was created between the previous block and the block where the hash is saved in. That's a starting point.
that does nothing to verify authenticity
it does something, sometimes. it pushes the required fabrication timeline back.
if it is mandated that every photo or video taken for the possible use in evidence is notarized at the time of acquisition, any fabrication would necessitate total premeditation. that is, the fabricators would need to know ahead of time what they were pursuing and what evidence they would need. this seems like a very costly barrier.
for example, altering security footage would require some fantastical elements: a real-time system of ingesting real footage and altering it in real-time to slip it into the notarization pipeline within the error margins.
requiring that any equipment that produces acceptable evidence stream commitment hashes in real-time to public append-only repositories would be an enormous step forward.
This sort of chain doesn't need PoW I take it, just a very secure police server to sign blocks.
And it couldn't be run by the police or any of their friends, since they're the adversary.
Might have a point. This was before blockchain.
I suspect that the cops wouldn’t like the chain public, though.
Like when people discuss voting, I believe a blockchain [0] is a terrible pitfall compared to a classic distributed database system of predefined nodes run by different organizations. For example, imagine a couple hundred predefined nodes run by different states, federal agencies, etc.
An attacker altering the ledger would still require compromising an unreasonably large number of independent groups at once, and even then the rest would be able to clearly see that some unusual and suspicious event occurred.
By limiting membership a bunch of problems simply vanish, like long-clearing times, wasting hardware on mining, vulnerability to foreign botnets, etc.
[0] A blockchain is distinguished by its core requirement, from which a cascade complexity flows: Uncontrolled node membership. Don't be fooled by people pitching "private blockchain", its a contradiction in terms designed to rehabilitate hype, like "multi-sample Theranos test" or a bicycle as "Segway passively stabilized inline wheel model."
"Crack the hash"? Does this mean you were employing some novel hashing algorithm and relying on its secrecy? If so your employer were never serious about security in the first place. Hardware attestation is more or less a solved problem, and that solution does not involve secret algorithms.
Eh. It was some kind of hash of the image. I was not involved in that project, so can't tell you exactly how it worked, but the images were "signed," and someone figured out how to "re-sign" an altered image.
I think it was a fairly well-known technique.
Now sell them version 2.
I imagine in this age of blockchains you could embed into a media file a signature that proved it was no older than the timestamp of when it occurred, the digital equivalent of a hostage-proof-of-life photo with a recent newspaper
But I don't know of a cryptographic mechanism to ensure that a digital image is not more recent than a particular time
> But I don't know of a cryptographic mechanism to ensure that a digital image is not more recent than a particular time
Many (most?) blockchain mechanisms include a timestamp in each transaction on the chain, so while multiple records from the same owner prove little (the timestamps could be faked over a given period of time) the interaction with the wider network and the chain would give some confidence that the record happened between within a small amount of time.
The other possibility, that doesn't require a chain with many independent active participants, is to have things signed by an external trusted authority. Submit a hash of the content and appropriate metadata to them, and have them sign it with a signing timestamp. I've considered abusing ACME certificates for document signing like that: the hash of content (or some signature based upon it) becomes the subdomain to sign¹ and you get a certificate that even after expiry is evidence that the CA saw that value at the signing timestamp. Note of the signing will also be in the public certificate transparency log. This wouldn't, on its own, prove anything about the authenticity of the content, that could have been doctored before signing, but it does prove that the content+metadata existed at that time (so might be more useful in copyright claim type cases, or agreed contract situations where all parties have signed the content and the signatures are included in the metadata, than for proving authenticity).
----------------
[1] based64²-ed with non-alphanumeric characters removed and truncated³ to fit or split, so acodha3sf7whsrhtqestkabtx0b4bbhyveee0ajnrpqcuxrjjvmhsujgcex.domain.tld or acodha3sf7whsrhtqestkabtx0b4bbhyveee0ajnrpqcuxrjjvmhsujgcex.w5jmmkpmyfgshx2jecsfordpnq.domain.tld
[2] names not being case-sensitive drops some of the entropy, if that is a concern use a 32-bits-per-character encoding instead and have names twice as long
Publish hash(image) on the blockchain at a verifiable time, then publish the image itself.
The image contains the previous block’s hash.
Wouldn’t this establish both a lower bound and an upper bound on the time the image could have been produced?
wouldn't that be a hash of the image signed by a trusted entity and stored on a chain? maybe i'm overlooking why this doesn't work
Interesting, There aren't any newspapers left in my country, neither printed nor not printed. The closest you can find is the weekly advertising booklet here and there. Which is irrelevant now because a computer can either stich new content to an old picture, or entirely producing a custom picture.
That would be a use case for a block chain. But I still don't understand how you are securing the integrity of the validity of the certificate stating the authenticity of the media. I only understand you are stamping media with a "at least as old as [timestamp]
If you want to prove that "happened at or after this timestamp" you can use a randomness beacon. NIST[0] and others publish a random number every N minutes. Embed that (or a combination) of those seeds to prove that you observed this value. This does not work for the harder problem of proving an event happened before a timestamp.
[0] https://csrc.nist.gov/projects/interoperable-randomness-beac...
Which country no longer has newspapers?
The end-game is that people will willingly surveil themselves 24/7 on behalf of The System because that will be the only way to prove what they didn't do.
Ah yes training the AI with more data to represent me even more accurately.
I suspect so. Tbh, I'm surprised it hasn't happened already with the amount of processing that cell phones do on photos, with generative fill/expand/perspective change, etc.
We are quickly going to reach a point where any photo or video taken on a smartphone is inadmissible by default.
You should see what people were capable of in the darkroom, let alone before all this. You could always manipulate imagery ever since there was imagery to manipulate.
This is why:
- the whole roll of negatives was prime evidence;
- police forces were one of the biggest users of Polaroid instant film.
And moreover, who had a darkroom and the skills to edit substantially a picture?
Whereas here we have nobodies being able to generate pixel-perfect fake "evidence" from the computers they already have.
Plenty of people. If you have running water, some tape, and trashbags, you too could have a darkroom.
https://www.metmuseum.org/exhibitions/objects/objects@exhibi...
The roll itself can be manipulated too. Most of the techniques used in modern photoshop are basically 1:1 carry overs of darkroom processes. Layers, dodge and burn, masking, etc.
There was a time you could take this class in highschool.
Big difference between that and writing an AI prompt.
Not really. End result is the same: manipulated image.
Are we really pretending like the effort to do something doesn't affect how often that thing occurs?
Are we acting like that was ever a limiting factor towards disseminating propaganda in the analog age?
We've gone from highly skilled people being able to forge some specific photos and documents using substantial time/energy/resources, to any asshole being able to generate realistic full-motion video in minutes.
I get that there is a certain type of moron who thinks that the collapse in cost of misinformation has no harm... but all you've done is announce to the world that you are a moron.
It is really not any different. People would throw a hubcap in the air and pitch it as a UFO photo and idiots would latch on to that. You could take a photo of the empire state building and use a double exposure to make it look like you were king kong. Kids were doing this sort of stuff. Stop motion home movies where you'd look like you were levitating or your head got cut off.
It always comes down to provenance.
How many people could do that?
How long did it take?
Now it’s a lot easier and faster
Compared to now
I’m still shocked we have not seen an extremely convincing AI video of a famous person or world leader announcing something huge like UBI or WW3 or aliens.
Surely it’s just a matter of time.
per ft.com: https://archive.fo/BIOej
[The Derbyshire Police] declined to give more detail
about what the evidential material consisted of.
The term [evidential material] can be used to
describe witness statements.I don't know if it's still the case in the UK, but in the common law and still in the US this why all substantive evidence, with very rare exception (e.g. dying statements), is witness testimony given on the stand. It may seem absurd when a witness or expert is given a transcript of an earlier statement or report just to recite it, but this is exactly why.
The loophole is all the powers the police and government have to more-or-less punish someone before a trial, or even before charges.
Maybe use the word “falsify”?
I wonder how many people have been unjustly imprisoned between planted evidence, made up evidence, and illegal parallel construction…
Here in the US? Probably a large double-digit percentage of cases imo…
> large double-digit percentage
This is a very very intense claim, and if true, would represent a monumental institutional failure across hundreds or even thousands of disparate organizations.
Do you have any data to support your hunch?
Strong claims require strong evidence.
When DNA matching was introduced, we discovered that at minimum 10% of people on death row were innocent. Death row cases are among the most litigated and examined cases. So, 10% is a reasonable floor, and we're already in double digits.
That stat is off by a couple orders of magnitude. The total number of death penalty convictions overturned by DNA evidence is 29 (as of 2025). There are a couple thousand death row inmates right now, and the denominator here is all the people who were on death row in the last 20+ years. That's a rate of significantly <1%.
https://deathpenaltyinfo.org/first-death-row-exoneration-inv...
Shouldn't the denominator be the number of people actually executed ? 691 in the last 20 years, for instance ?
it doesn't matter if it's 29 or 2900. Even 1 is wrong.
The commenter isn't litigating that claim, they are litigating the claim that at least 1 out of 10 of those on death row were false.
"we discovered that at minimum 10% of people on death row were innocent"
How did we do that? I never heard this: certainly 10% of people on death row weren't exonerated by DNA? This is some kind of shaky extrapolation I assume?
> This reasonably sets a floor
I disagree wrt reasonableness. It’s just too big a leap. There are a lot of crimes, and not many land you on death row.
The observation was that death row represents the highest level of scrutiny, and still had 10% false positives for guilt.
Is there any argument that less-scrutinized cases would have a lower level of false convictions?
The 10% claim has been refuted.
Hoo boy, welcome to the history of the United states.
Shouldn't it be the exact opposite here ? The burden of proof is the other way around.
The big claim is here: the state has grandiose claims that the overwhelming majority is fair, but there is no proof of it.
Therefore the state should prove that more than 90% of the cases are legitimate, fair, not coerced, and not motivated by the pressure to interrupt the proceedings.
97% of people choose plea deals or out-of-court settlement, it is a huge amount.
It means that in real practice, not imaginary internet, people who face court consider that justice is a big machine that can crush you no matter if you are innocent or not.
In the best case you are acquitted at the end, but you are guaranteed to bear the financial burden, fear and stress as a punishment.
Being held in jail before trial is a very convincing reason to plea deal too.
It's a system engineered to make pleading the only reasonable option, no matter if you did anything or not.
That is true--the checks and balances the founding fathers fought so hard for were thrown out the window with overlegislation and expansion of prosecutorial discretion in 20th century. To make a convincing argument that "double digits" of cases involve fabricated evidence, you still need to explain why prosecutors would engage in fraud at this massive scale. Just laziness? Collecting scalps? The incentives run that way in some limited cases, e.g., prosecutor up for election, post-reconstruction south. But you need some explanation there.
They get rewarded based on winning cases?
Yeah, again, there are some incentives to fabricate evidence like career advancement. Now why should those, on a mass scale, outweigh disincentives like getting caught in an adversarial process and (presumably) some qualm about regularly convicting innocents and regularly letting guilty parties run free in communities. Easy to argue in particular cases but I haven't heard the basis for a trend.
> Shouldn't it be the exact opposite here ? The burden of proof is the other way around
That's the rule for criminal court in the US, but each of us is free to pick his own standard for his own purposes.
A burden of proof is associated with an individual claim. There’s no “burden of proof in the other direction” - what you’ve actually done is created a second burden of proof and also - worse - attempted to distract from the original point.
It is disingenuous to weasel out of proving one claim by making another, or saying “look over here”
Also, outrageous claims in opposite directions can both be bullshit.
On what basis is it an outrageous claim? You think the number is closer to 0? That sounds like a more outrageous claim to me.
That is like claiming that double digit percentage of software bugs and vulnerabilities were intentionally put there by malicious software engineers. Its outrageous to claim its that high.
Even single digit percent is hard to believe, but its possible, but double digits you are talking China or Russia levels of state corruption and even there I doubt its that high.
~~Please point to the place where I said your claim was outrageous.~~
Edit; upon closer examination. I did imply in my last paragraph that your claim was outrageous. Bit of a gaffe considering I’m the agitator here. My apologies.
A few years ago, one of my coworkers was arrested for a domestic violence complaint. Looking into his case, I found an extremely specific lurid description of the allegations -- and then I found the same lurid description copy/pasted to every other person recently arrested for the same crime. I'm probably getting the specific terms wrong, but I did click through to see it on a government website, because my first suspicion was the aggregator, but no, the police just had a boilerplate story full of specifics which could not possibly apply to each and every person they carelessly slapped it onto. This absolutely blew my mind at the time, but it fits with smaller subsequent observations. In any case: a double digit percentage of institutional failure does not upset my priors about how carefully the police operate.
Can you put some of the text in a comment here?
Why did you "look into the case of your coworker"?
You wouldn't be curious?
An important thing you should recognize: the judicial system is painfully nontransparent in such a way that even figuring this sort of thing takes an extensive amount of time and is often even impossible. I've personally gone down a similar route (did some journalism for a bit) by trying to understand how shotspotter is used in prosecution, many of which resulted in false arrests and many, many years of life lost across all the people arrested falsely from it.
If you would like to begin trying to answer these, I recommend starting with submitting some FOIAs. Considering your stance seems to be that you won't believe what others are telling you -- I promise you that you'll be surprised.
If you believe parallel construction should be illegal (it sure seems like it is unconstitutional to me), then 100% of prosecutions that rely on it are unjust. I don't think anyone truly knows how common it is, though, and that's by design. Double-digits wouldn't shock me at all.
Police in the United States is already in a state of "institutional failure"...
“Police in the United States” is not a monolith.
It’s easy to say things that sound true on the surface, but even if true, it’s still irresponsible to say them on the back of a hunch.
It's more monolithic than you would think due to shared culture over the internet. There's a whole narrative about sheepdogs (them), sheep (us) and wolves (the bad guys).
I don’t know the numbers, but DNA exonerations give a bit of a natural experiment (where testable evidence was preserved).
They give a floor, and that floor is too small to be useful.
Do you have any exposure to the criminal "justice" system in the USA?
We have the highest proportion of imprisoned citizens in the world.
This is done because there's an exception in our constitution for slavery "as punishment for a crime" and well all know that capitalism loves slave labor.
Especially if law enforcement uses Parallel Construction [0], lying to the court about the process taken.
Consider all of that can be used for forced confessions and forced plea bargains also. In those cases, the "evidence" doesn't even need to exist at all, or be on the record in any way.
Let alone all it takes is a photo and a voice clip of 10sec to create an imitation of that person confessing to whatever your heart desires.
An old phone book is enough.
Hows wolfie doing?
it's max
aren't you an imposter ?
Sadly, there's more evil and more laziness/incompetence in the world that's being accelerated by AI than there is good.
Over all time? Probably tens of millions.
Can we know the motivation? Will it get them a bonus at the end of the year? Was it something common in the cases, maybe similar victims or something else?
I'd wager that it was just a shortcut to getting his work done. That banal motive is why we've seen an explosion of these cases and why they won't stop.