The SOLUTION to FAKE AI pictures & video

Go to squarespace.com... & save 10% off your first website or domain with code “Tony"
Tony Northrup describes the two-sided problem of fake AI images & video. The first side is obvious: AI pictures and video are constantly used to mislead and scam people. The second side is that real images and video are being falsely declared as AI so they can be dismissed.
Today, social media is trying to fight this problem by indicating which pictures and video are AI-generated, but they're doing a TERRIBLE job at it. Most fake AI images can't be detected, and all efforts at labelling them are ineffective against bad actors and scammers.
The better solution is to label REAL images, but that's technically complex. Tony Northrup lays out a workable solution using public key cryptography and secure image processing hardware that can eventually give us green verification checkmarks on images that ARE actually real. People will naturally begin to question any image that isn't verified.

[@noriller]

The biggest reason this might not work is: people just don't care. You can show them irrefutable proof, but if it's not what they already believe in... they just won't care.

[@daran0815]

Even if they care, they'll excuse any amount of proof as being exceptions.

[@Wissperwind]

If you buy a 4k TV, display a AI generated photo, take a photo of the screen with your TIM camera, size it down to social media size an ... BAM, you have a valid AI generated image. Seems to be a new business idea...

[@SatongiFilms]

TPM can be spoofed, pretty sure TIM would be too. All software can be hacked.

[@bab008]

FB feed is loaded with AI generated "nature" photos under the name "National Geographic something or other" to name just one. And I see friends sharing and liking what they think are real photos or videos of birds or other exotic (and impossible!) creatures in the past several months.

[@Hodenkat]

And worse, many of those images have comments that suggest some think they're real. 😞

[@lyfandeth]

Well, as Dr. Land said 60 odd years ago, if it isn't a Polaroid, you just can't trust it.

[@morvegil]

Who's that

[@fepatton]

@@morvegil Cofounder of Polaroid and a key inventor of the practical instant camera and film. Instant photos were very difficult to retouch, so in theory more trustworthy.

[@Bugside]

Nowadays there's instant photo printers, so not even that

[@tubularificationed]

OF COURSE you can fake with Polaroid. Just use it to photograph another picture (or even simpler, a computer monitor screen with sufficient resolution) 😉 That's why Tony demands, that digital cameras would require to include - focusing distance information, and/or - stereoscopic information from secondary cameras on smartphones, and/or - a video fragment in their TIM-protected metadata, to combat this type of cheating (else, also digital cameras would be vulnerable to such a cheat). Having said that, the - focusing distance information is a too-weak protection? Just screw a close-up lens 'filter' in front of your optics (preferably of a high-quality 'achromatic' type), then the camera-recorded focusing distance might record 'infinity' whereas it actually sees e.g. a smartphone's screen. But as Tony said, EVERY technical measure which makes cheating - more difficult, - more work, or - more expensive, makes it less ubiquitous, and is therefore better than not having it 🙂

[@YiKidane]

kahma AI fixes this (AI Headshots). SOLUTION to FAKE AI images.

[@STOVL93]

I like the analogy of locks on a door. A solution doesn’t have to be perfect, it just has to make spoofing more resource-intensive for a bad actor. I’m really glad that you’re talking about this!

[@RioHelmi]

As a photographer with a 44 year career behind me (hey i am still active), two things you said: "people want to connect to real people" and "everybody is sick of perfect pictures" (more or less what you said). I have been shooting digital since 95. Do I want to see TIM implemented? Absolutely. I can pick an AI generated picture most of the time, but it's getting slicker. AI is one of the most destructive forces for human creativity.😢

[@TechLeatherCraft]

But... Do you remember when shooting film, and digital started to gain ground. The forums were filled with people talking about the downfall of photography due to digital. I suppose time will tell about this Ai stuff. (to be fair, I'm not a fan of AI stuff either... But I'm old 😂)

[@bngr_bngr]

@@TechLeatherCraftnot like film was more real than digital. No picture taken by Ansel Adams was ever real.

[@offgridrvliving]

Or is it? What about the blind, who can use AI to be creative for the sited, or the hearing impaired who can use AI imagery for communication. And so many other uses to aid the less fortunate. Using human intelligence, we can use AI for the betterment of humanity in so many ways.... Interested in your thoughts.

[@captinktm]

I agree, it's gullible people which are the problem, we old un's use the experience we have gained from living without the web. Anyone under 40 gets the experience straight from the internet. It's a bit like us getting all the latest info down the pub from the pub drunk.

[@houserhythm]

How is AI destructive for creativity? It literally is a new creative medium. You can have your words made into images. Yes, it's destructive for authenticating photography, but by itself it's an amazing creative tool - it allows people to be creative in a way they couldn't have been before.

[@riyaansheikh7470]

Plot twist: This video was AI generated

[@stevemuzak8526]

This post was AI generated. You don’t exist

[@Black_Jesus3005]

@@stevemuzak8526*suprised pikachu*

[@davidb9670]

Who cares if AI destroys any last shreds of trust left in media? Let's go back to seeing things with our own eyes. Let's go back to experiencing people face-to-face. Let's go back to reality. Let's go back to being human. There are a bazillion beautiful photos out there. There are a bazillion interesting photos out there. AI is not going to flood the market. It's already flooded. As an amateur photography, the only photos that mean anything to me are my own, and that is because they are mementos of my own real experiences. Moreover, for me at least, photography is more about seeing beauty in the world, than it is capturing images. I don't even need a camera. Some of my best photos are only in my head. AI will never touch this.

[@Grassdia]

EXACTLY, this is why I always argue that AI may have the ability but it doesn’t have the intent and artisan connection to the piece so why would anyone else.

[@STOVL93]

I agree with you, but I think Tony brought up a couple of good points that don’t have anything to do with artistry: What if a someone uses AI to impersonate a loved one in distress and demand a ransom? What about when bad actors use fake journalistic images to influence public opinion? Not every picture you take for yourself would need this technology-obviously if you shoot on film it would be impossible, but there are certain situations where we would want the technology Tony’s proposing.

[@flickwtchr]

Just because you deny a problem exists doesn't make it go away.

[@testcams]

The solution you're proposing is already part of the Content Authenticity Initiative, which includes provisions for device-specific cryptographic keys that are used to digitally encrypt and sign images coming off the sensor. Note there are various methods to implement this, some which wont require a TPM. For example, every sensor has a unique fixed-pattern noise pattern that can be used generate the key - the full FPN data is only available to the hardware and firmware so there is some built-in access protection similar to what a TPM provides. The CAI then extends this concept throughout the workflow include edits, using cryptographic-signatures on the edits/modifications.

[@DJVARAO]

Overcomplicating metadata won’t deter bad actors. Not sharing your photos, travel logs, income level, and lifestyle will. People should just stop sharing their private lives on public websites.

[@Black_Jesus3005]

Agreed. I stopped sharing photos of my children online a long time ago. If my family wants to know how I’m doing or my kids they can call.

[@andrewclifton429]

That's not going to happen at scale, so it's a pointless suggestion. So long as most people you know share some of their images with some other people, they're exposed to bad actors and therefore, you too are vulnerable to AI fakery impersonating them. To suggest we can somehow collectively avoid this problem by "not sharing images" is a bit like saying we can avoid the hazards of printed propaganda by all agreeing to quit this new-fangled "reading" business!

[@batuhancokmar7330]

So you protect yourself from getting fakes of your own photos. That is one (and not the only) solution to the one half of the problem. Other half of the problem is; how do you know photos you see is real... This IMHO worries me more; because I can know for certain if my photos are faked. But I can't know for certain photos/videos I see are not faked.

[@unstanic]

@@batuhancokmar7330 With his suggestion it wouldn't matter if they are real or fake. Could be real, could be fake, who cares. The problem is when fake images do harm. But as @andrewclifton429 mentioned this does not apply to propaganda that we consume. I think for the latter, unfortunately we will have to assume an image is fake till proven otherwise :(

[@mikldude9376]

The problem is money , as long as there is money in some way involved with anything in life , corruption abounds . And secondly what is the incentive for companies to, journalists , camera makers , to big tech companies to implement some form of security to pictures ? Unless there’s some payday in it for them , it’s hard to see them doing it .

[@redriverpost]

Problem is the overwhelming majority of people prefer fake.

[@kylegoodman5196]

I don't think so.

[@contentm3893]

@@kylegoodman5196 Sure they do because they don't care how it was created. AI uses artist copyrighted material to learn from and that includes photographers.

[@kylegoodman5196]

@@contentm3893 Highly doubt the overwhelming majority of people prefer it.

[@contentm3893]

@@kylegoodman5196 It is when the final result looks just like professional photography or video. AI is impressive tech.

[@kylegoodman5196]

@@contentm3893 It doesn't though.

[@adamginsburg9909]

The solution is don’t use Abode software to edit your files, because they steal every image you edit in their programs and upload it to their cloud server and feed it into the algorithm

[@ProfoundProductions]

Too late, Ai already has all the data it needs to create anything at this point.

[@RandumbTech]

It’s hopeless. Best to just accept that we can’t stop this crazy train.

[@STOVL93]

I got the opposite out of this video. Tony gave me a lot of hope for how a practical system might look in the future!

[@KNZ5]

It's not about stopping it. Fake images will always exist. It's about making it harder to create fake images.

[@RamsesTheFourth]

@@KNZ5 Its always going to be easy to create fake images.

[@KNZ5]

​@@RamsesTheFourth I’m a software engineer, so I have some experience with how easy software needs to be in order to be accessible for most people. It has always been possible to fake images, and it always will be. However, if the software (from the major players) that generates AI images is required to mark them, this could reduce the number of AI-generated images that pretend to be real by about 90%.

[@RamsesTheFourth]

@@KNZ5 That would be better option than what Tony is suggesting. But still, real images can be marked as AI generated. And AI generated images can get the marks removed. It depends on how it is implemented of course.

[@Kellysher]

AI will change a lot of things, eventually. We are visual people. Even if you think or know something is AI, you can’t unsee something. Your favorite celebrity killing a puppy, or assaulting a woman. It won’t matter if it’s marked AI. You will always remember what you saw and somewhere in your brain, it will create a bias towards that person. People are already easily manipulated by bad actors. These tools will make the bad guys 1000x more dangerous. I would hope that we try to regulate AI soon, but my confidence in that is pretty low. Glad I’m in my sixties!

[@GemmaHentsch]

My biggest concern is that it doesn’t seem engineered in a ‘the company cannot fake the image’ way… as in it needs to be secure from everybody or it’s not really secure.

[@annoholics]

So you want a system that can label photos that as authentic. It still can't detect authentic photos that are not labeled in any way. There are billions of these photo's. The social media platforms have no interest in labeling photos because the majority of their "customers" don't have camera's that authenticate photos. This is a similar problem as spam. If everybody would digitally sign their emails as original then spammers would not have a chance, right? How many people use a digital certificate to sign their email?

[@houserhythm]

The customers of social media platforms are advertisers. Users are their product. They sell your attention to advertisers.

[@annoholics]

@@houserhythm Did you notice the quotations around the word "customers"? That means they are not really customers, although I am a paying customer of KZbin and I don't see any advertisement from KZbin.

[@brohanson]

so if i had a little blog of my own I have to maintain a certificate system to display images as genuine? how does displaying the genuine symbol work? where does the certification come from? do sites have to display this information or is it built into the web browser? how does revocation work? what if my photos aren't online for years at a time?

[@wanderlust0120]

Solution : stop using social media. Then, if possible, get a life.

[@KaiTiura]

What a genius you are...

[@zulfika_]

"said by the guy on social media"

[@wanderlust0120]

@@zulfika_ I don't use any social media. Just KZbin for tutorials and WhatsApp for messaging. But I have to admit I don't have a life 😭

[@RamsesTheFourth]

@@zulfika_ If you scream it out of your window hardly anyone would nottice.

[@JuneisyHawkins]

@@wanderlust0120 So you DO use social media.

[@seraphin01]

I'm sorry I don't see any of those things fixing the issues.. First of all, I edit my photos, I remove dust marks off the lens, pimples, stray hairs, sometimes fix bad make up, whatever you name it. Of course not on every photo but depending on the subject and client and all it happens. It's not AI whatsoever but it's edited. With those changes my photos would end up next to crappy ai generated images.. Some would say it's fair.. But what about film altered images? They're still altered even though it's on film, those get a pass? Worse yet (imo) : staged photos. As a nature enthusiast, I love taking birds photos, no cheating, going out, stalking birds and taking photos in their natural habitat, just cropping and color correcting the final image.. And at the same time some photographer just stage photos with feeders, cabins, fake branches, etc.. Or straight up use tamed animals or in captivity.. And those get to pass as natural photos That's like double standard, when ai would get flagged, but "fake" images passing as totally normal. Anyway I think none of those solutions will come to pass, it will be like it is today with photoshop portraits or staged nature photos, people will think it's the real thing and nobody actually cares in the general population

[@comeraczy2483]

You are absolutely right. Not only it isn't clear what the problem this "solution" is supposed to fix, but it will have nefarious unintended consequences. I already imagine the devastated newlywed couples whose favorite wedding photographs are labeled as "AI generated" because the photographer had to edit out something unsightly. Even more tragic: children whose parents just died can't use their favorite pictures to remember their loved ones because it is labeled as "AI generated". Editing important and meaningful pictures is an absolute necessity and IMO an important civil liberty. The idea of inviting governments and large corporations to deprive us of this fundamental right is abject and revolting.

[@seraphin01]

@@comeraczy2483 I didn't think about those possibilities but it's indeed very troubling.. I wouldn't want my important photos to be tagged as "AI" like I'm some freaking pathological liar.. ugh

[@Bladeclaw00100]

This is a really good solution. Now the challenge will be to have everyone buy a new camera since the existing ones do not have a TIM chip. This will be very difficult for companies and individuals that have invested lots of money on a current camera they are using. They would have to repurchase their entire camera inventory for this one feature. And I bet there will be some people who will out right refuse to give up their old camera but still take real but not validated pictures. It's just like phones or TV's, where analog signals are out and digital signals are in, so everyone needs to switch to keep up with the new times. The problem is this will take time for people to switch especially for people who more recently bought new cameras. During this transition time is where we will be most susceptible to all the vulnerabilities. But, I can see it improving over time as more people switch over. Great Job hope you get royalties for this.

[@MrsGypsumFantastic]

I’m not a tech expert, but you can now buy a WiFi charging chip for phones that didn’t have wireless charging capacity. Could something similar work here or is this type of technology that’s too integral to be added post manufacturer?

[@bobbullethalf]

Expensive cameras and lenses will make photo and video obsolete. I can just type up a few phrases and have an 8x10 of any photo on my desk. No need to pay for an expensive photographer or purchase an expensive camera.

[@d_s_spence]

I agree that this is a huge problem. I had not thought about it as far as all the social media mess, but I want to have just the first part for myself: a camera that digitally signs any photos that I take. That is enough to give me some peace of mind. But cameras that do that are expensive. I want that to be the norm. I am okay if the digital signature is invalid after any edit. I just want to be able to prove than an image came from my camera and was not modified. I keep the original unmodified versions of anything I do modify and "publish" on social media anyway. This would be like being able to produce the negatives if asked. Good work on thinking thorough a lot more of this. I think your ideas are valuable and definitely going in the right direction.

[@Kliffot]

Really like that idea !

[@Teslien]

After discussing with people, we've come to the conclusion that everything should be labeled as: AI altered, cannot be determined as AI altered and no AI alteration if given enough information. All images and videos will have to be categorized as one of these 3 labels. If someone wants to show off AI skills, sure they'll appropriately label. But the majority of AI creators do not want to do that because it'll hurt their use platform standings

[@leokretzers8883]

Great approach, Tony. As an engineer I like the idea that the tech created this so tech should resolve it (within reason, lock on the door). Although tech companies like apple should built this, still public bodies, government, should run campaigns to educate the people that they should want this. Like the ssl encryption for their banking servers. I’ll try to think of some quirks like you asked but for now, great you dare to put up such a ‘Ted’ talk specific content to your 1,63 million users. It rocks the boat, men, most folks don’t see this shit coming to them. Please continue, might be one of the most important things you two have been doing in years. And I bought a couple of your books, so know I value your work, but this is instrumental for personal safety. Heck, let’s go big, it is instrumental part for future democracy.

[@brandondaniels9471]

Plot twist: it really wasn’t even Tony in this video. It was really just AI.

[@Nam3Iess]

I am in a beginner photography group on Facebook and the over use of the photoshop AI stuff is crazy. Everyone wants "perfect" pictures and they happily do body modifications as well 🤮 This should be called compositography or something like that...

[@RetrieverTrainingAlone]

Is denoise AI? Is cropping AI? Is dodge/burn AI? A slippery slope...what is "significantly edited" image?

[@carultch]

No. All of that is deterministic computing.

[@RetrieverTrainingAlone]

@@carultch Topaz AI denoise, Lightroom automatic subject selection, ON1 Photo RAW Brillance AI

[@contentm3893]

@@carultch But AI is going to be integrated into the software we used very soon.

[@ghostreaper1444]

Terrible idea. NSA will be so happy if this implemented. Will be so easy to identify the source of a photo. This can be misused in so many ways. Privacy will be completely gone.

[@scb2scb2]

Its a myth that it has to be like that sure there can be issues but the 3 letter agency's have been against making systems more secure (in open methods) for decades from the fight over pgp, ssl, locked hardware it has made their job harder not easier... Just wait until you see the lobby against this in politics as proof it won't be a step forward if done correctly. Now you will say but they only do that for 'show' they got a deal with say apple and a backdoor thats why public review is needed but i doubt that can even happen in the first phase of this for all kind of reasons.

[@comeraczy2483]

Even without the NSA it's a terrible idea. Do you want to see your favorite pictures of you your lost loved ones labeled as "Fake: AI generated" because the photographer had to edit something unsightly out of the frame?

[@scb2scb2]

@@comeraczy2483 Watch the video its not about not being able to edit but doing that in a way that the viewer can see in the meta-data that it has been edit in a way that can't be removed without breaking the seal. There are technical issues and edge cases for sure but his goal/idea is not to stop editing but to create a clear paper trail included in the image what has been done and with what kind of tools.

[@comeraczy2483]

@@scb2scb2 ​ Thanks a lot for you extremely helpful suggestion to watch the video. Who was suggesting that the video was about not being able to edit? My comment was explicitly about people editing images in the context of tracking edits, particularly in the context of the claim at 4:40: "we need a system that validates a real image has not been significantly modified since it was taken with your camera...". The "significantly modified" part will be a matter of arbitrary judgement, and there will be important and irreplaceable pictures that will be labeled as "significantly modified" for arbitrary reasons. I don't think that many people would want orange question marks or red crosses on the pictures that they use on a FB merorial page for a loved one, or on their wedding page. For some, it could be devastating.

[@scb2scb2]

@@comeraczy2483 The way i got it was not to limit how much it can be edit but when it looses its status as 'real'. My first comment also reflected that to Tony. Its not a easy problem at no point should we be stopped from editing and personally at this moment in time we can add a few % in size and add a 1 bit layer that signals what parts have been 'really' changed. I would even not be against that a unedited (say mid resolution) should always be included so the end user can decide when the system labels a image as 'not real' enough. But its not a easy question the real question is should we start thinking about this or just accept we can never trust a image again or do we try to at least do something to add meta-data on what has happened so users can judge.

[@NatanielsArt]

Definitely an issue, all the ai garbage and I haven't seen even 1 of them labeled. For now I nightshade and glaze my photos, so they can't use it. That may be good solution if it could somehow be implemented to existing phones and cameras. Important for journalism but as a landscape photographer I'm jot a fan of having gps on. What about editing, if I use 3+ softwares for astro, NR and focus stack would it still recognize a photo?

[@cfoote416]

It's a marvelous idea I would consider Using the blockchain to offload some work and enable the ability to do extensive editing

[@Original_Old_Farmer]

I do appreciate many of your videos but I think you have a bit of what I call Apple-think mindset which makes little sense to many of us. Granted you want a way to proof photos and videos. If it is news, I can see it or even for competition. Most everything else is artistic and really doesn't matter. It's the image people like. The examples you offer to solve the solution is a good thinking through of the process. You seem to insist upon location as part the process. If I take an image at my home, I don't want location. If I am a professional, and technically I am, it may be a location I do not want others to find. I've left the GPS turned off on my camera. In fact I have not even set it up. As for cropping and other components of an image, there may be reasons not to show what has been cropped out. You've mentioned many reasons in past videos dealing with other topics. The cost of the process in a camera would mean only Elon Musk could afford one and it might need a fork lift to move it around. Also, I don't use Photoshop or Lightroom. Never have: never will. If the software I use added such features, I use a Linux computer, I would leave it turned off. I don't want to be disrespectful, but if you don't want my product because these things are not included, so what? I really don't care. AI will be a part of photography for some people, others won't use it. It comes down to the buyer's personal choice and whether the person taking the picture is honest. Maybe you care, maybe you don't. In theory, if I can make a hamburger more appetizing using AI, I will. So what? The majority of individuals for most purposes just won't care. Good video about theory.

[@JaroAtry]

The problem is that most people who fall for these images don't care about the evidence if it's real or not.

[@half-decentphotography6544]

Technically speaking, this is a great idea - in theory. Convincing tech giants like Apple and Google to collaborate on it, and governments to support it, will be very difficult indeed. These institutions are oligarchies, owned and operated by the kind people who are normally totally focussed on their own narrow self-interest; it's wildly naïve to expect them just to "do the right thing", because they should. The best chance of success is to convince them that widespread abuses of AI threaten their selfish interests - and genuinely effective countermeasures will be good for business. If you can do that, it may actually happen!

[@unstanic]

Especially when the interests of these guys is to actually share fake photos. Facebook has been involved in political propaganda before, it's not like they minded it.

[@contentm3893]

What is the wide spread abuse actually? Can you name one photographer that has been damaged from AI? The copyright violations with Ai has already happened with artist and nothing has ben done to stop. AI. Tony is just looking to benefit from AI and make money without caring about the artist copyrights that have already been stolen. The regulation for AI should be that it cannot use anyones copyrighted material, likeness or IP. That's how you control AI. If you or Tony cared you would be upset every time a Guns N Roses Sony or music video is posted on IG or TikTok with out permission. You and Tony have become the bad actors who are out to make a buck and say that you are protecting everyone from AI. Well if no one will protect music artist and the people that created content for them back in the day why do you think someone's wildlife photos from Yellowstone will be protected? You guys are not fixing the problem but just becoming a used car salesman trying to make a buck.

[@comeraczy2483]

Is that the problem that this video is trying to solve: preventing the widespread abuses of AI? If so, first, is there evidence that there is widespread abuse of AI? Then, if there were such evidence, wouldn't it be a lot better to resolve specific problems rather than unspecified "abuses of AI"? If there is a problem, why should we care if AI is involved or not? For instance, tech support (and other) scams are very successful without AI, so why bother addressing the specific problem of tech support scams that use AI while not much is done about all those scams that don't use AI?

[@half-decentphotography6544]

@@comeraczy2483 There are many different applications of AI - and many of them are no great cause for concern. I'm not especially worried about people using AI to insert dinosaurs into family photos! What Tony talks about in the video, primarily is deep fake technology - and there is already ample evidence of abuses in various domains from dating site frauds to political misinformation, revenge porn fakery and blackmail scams. More generally, if you develop tech that can be used to create extremely convincing fake images video and audio of any human being - and make that tech readily available to anyone who wants it - it should be blindingly obvious that it's going to be used to abuse, defraud, deceive and exploit people... UNLESS, you also develop some means of verifying that digital content is for real, and not fake. That's what Tony proposes.

[@contentm3893]

@@comeraczy2483 Because Tony is trying to make money by creating software that no one needs but he knows he can scare people into having to buy his software. The problem is with AI and how it's not regulated.

[@rschellie]

You could use something that works the same way as IFF (identification friend or foe) used by the military. When a digital picture is opened it sends out a challenge over the internet that compares the image being viewed to the original version that the photographer has encrypted and placed in an online server. If they match, then it is validated as an actual photo, if it doesn’t match, it is flagged as a fake.

[@TomHofmann]

Forced GPS location is a terrible idea.

[@scb2scb2]

He started optional i am also not a big fan of the idea but AI could be used to detect the place and provide the info to the viewer anyway (and how the place really looks) this also opens a danger. For example i might see a street outside a window and AI can find the place for me and i still know her/his location and become a perv.. This is something we have to deal with anyway sorry to say even if no GPS is provided.

[@TomHofmann]

The problem is that the vast number of people wouldn't know it's optional and why they should turn it off. The street sign example apparently is already happening... at least that's what a federal police officer mentioned in a recent talk on cyber safety I attended.

[@scb2scb2]

@@TomHofmann Yes default should be off but like you stated its a problem anyway adding several parts of meta info or data from the content itself can be very powerful and there is no easy fix. In some ways AI changes might help to 'hide' the location but that opens a whole new can of worms. This is what we get if we connect a whole world with meta-data and visual data over a global network in a way we are 'evolving' and need to adapt to this new reality somehow.

[@Hodenkat]

It always pays to have that one thing that only you and a loved one know about. A test. "Remember the name of that dog you had when you were 10 years old?" It could help you from getting scammed one day. I've seen some pretty poor AI photos that people commented on in a way that suggests they thought it was real. It always gave me a giggle. Now, I'm more concerned. AI is fun to "play with", but most of it is easy to spot. At least for me.

[@contentm3893]

This will never get off the ground bud. Listing you talk through this video and your case studies, this would be information overload for the majority of the public and you missed a key talking point. The green check marks are verification of real non AI images. But what you missed is if people don't sign up and use TIM then their photos and videos that are real will be accused of being fake which goes against what you are fighting for. This next level of control would be that no images can be posted on Meta unless it's verified as a real image. The point is AI content needs to be checked and verified as AI. That's it. That's the regulation all AI needs to confirm to and follow. AI regulation.

[@flickwtchr]

I'm annoyed that we are in this situation to begin with. Yes, we have to deal with it, but the bigger issue is that these tech companies have become way too powerful and greedy without any concern whatsoever how damaging their tech can be to society at large. Their hubris is just astounding.

[@matthewbary1]

I agree that something should be done to help prevent all of the fake photo issues and some sort of authentication does make sense but it also causes issues in other areas. Several people myself included use lots of vintage lenses on modern cameras, the adapters are not going to be able to transmit lens type, focal length, aperture setting or focal distance which could invalidate the photo. Also several people starting up will buy older digital cameras for cost and also several people buy older digital cameras simply for the fact that they take great photos and do not require more resources to be used as they are being given new life, these older cameras would not be able to be properly authenticated as there would be now way integrate a chip for them. Yet another issue I would see is film, you could argue that film already takes care of this issue being a physical medium but suppose someone took a picture of an AI image on a 4K or 8K screen. Then there are film scans, how would you authenticate a digital scan of a film negative. Again I am not knocking this idea as I believe it is very important I just do not know if an authentication chip is going to be the best solution and it would then cause all manufacturers to push out ANOTHER camera and possibly lenses that you would have to buy with the authentication technology .

[@ChrisLivingston33]

Uh... This video really crystallized how hopeless it really is. Everything will be questionable going forward.

[@valkrys68]

The answer to AI in photography is to use film by the newsies that follow politicians and nature. You can manipulate dodging and burning, but the ability to fake a photo is very difficult with films from Kodak and Ilford.

[@ronpierson]

If an idea like this were to be successfully implemented I feel like it would likely be short lived. We are on the precipice of a technological change that we can't even comprehend at the moment. Not only with the exponential growth and realistic capabilities of AI, but also with that of Quantum computing. In the same way we can rent time on various powerful AI computers at this very moment, we will also be able to rent time on Quantum computers. And with that comes massive processing power that has the potential capability of defeating most forms of current cryptography; rendering passwords, cryptographic keys, and security certificates virtually useless. And this is not something that's far off into the future. A search for something like "Qubits for Hire" will show that this is beginning right now. It's difficult to imagine what our very near future is going to look like as the capabilities of advanced AI meet those of Quantum computing.

[@rpmorrisjr]

The answer is photographers need to stop trying to beat AI at its own game. The more photographers move toward highly processed photos the more they’re moving right into the wheelhouse of AI. They will not win that fight and yet they continue pressing toward it. Return to photos that look like humans took them with a real camera, and AI will not replicate that. Don’t believe me? Ask AI to produce something in the style of Peter Lik. That’s child’s play for AI. Now ask it to produce something in the style of Cartier-Bresson and see what you get. See what I mean? You (pro photographers) are playing into AI hands. You’re surrendering to an enemy you can’t beat. I have no idea why you’re choosing to do that.

[@DrClementShimizu]

Social media has the opportunity to create a line of trust between who creates content and who views it. However social media companies allow spammers and bots to run wild.

[@notallaboutmeministry3285]

People pick the softer, real, photograph often instead of super sharp.

[@untouchable360x]

“There are no solutions. Only trade-offs.” Thomas Sowell

[@paulhiggins5165]

I'm not sure that the commercial users of photography will really want AI images to be clearly labled because this will make using AI Images in place of photo's more problematic. It's one thing to use an AI image to save costs when your audience don't really know about it- it's another to be forced to label that AI image as fake, which undermines it's value. The only way to address this issue is to ban any image generator that is capable of producing fake photo style images- because there's only one reason for this capability to exist, which is to produce images that look photo's but are not photo's. This an inherently deceptive capability. But-for the reasons given above- no such ban will of course be implemented. It's important to remember in these debates the entire value of an AI generated 'photograph' is in it's ability to 'pass' as the real thing- to fool people into thinking it's real. So I think that any attempt to force AI images to carry labels will be resisted by the big commercial users of photography in the Advertising Industry because they want to replace expensive photo shoots with AI generated images, and being forced to label those images as fake would damage their campaigns.

[@TheDroneAngle]

I'm one of those evil people that shoot real estate. Clients do not want realism. They want perfect lawns and paint, no concrete cracks, even fake skies.

[@comeraczy2483]

If you were shooting people, it would be just the same. Moist of us don't want to remember our lost loved ones with unsightly things in the frame. We don't want these pictures to be labeled as "Fake: AI generated" either.

[@Kristianpont]

Its a very real problem, and the solution needs to be simple and effective. Allowing some edits and not others will quickly turn into a mess. In my view, it can only be the basic image coming directly from the sensor that needs to be digitally signed via the TIM. Storing this image (it does not need to be full resolution) along with metadata on time and exposure, as well as a cryptographic signature, will serve as proof of authenticity. Changing even one pixel value will break the validation. When presenting your final edits, you can refer to this image as proof of authenticity, and it will be up to the viewer to judge if it has been modified too much

[@ihateunicorns867]

Me sat here loading Portra 400 into my Pentax 67: "Ummm…"

[@thomashenden71]

In this regard - the retro trend about using real film with visible grain is convenient, though it is just part of the solution. It seems like watermark technology that verifies that the image is not altered, from it was taken, by a real camera, other than contrast, colour adjustments and normal things. It would be completely possibly to make an image, that is watermarked by the original sensor data, verified to have been taken by a real camera and a limited set of normal modifications done. The watermark should also contain encrypted data about how much the image is modified, again - only contrast, w/b, colours, and the normal things. Am aware that this could limit the photographer to not use too much retouching and other minor modifications, however more important things are on stake, like the realiability of news photos. It should even be possible to have verification of the fingerprint of already existing cameras / sensors, which may not show up in a compressed file format, however would show up in a raw file. I cannot imagine AI being able to fake raw files, at least not yet, meaning - every journalist should be prepared to present the original raw files, to verify authenticity of the images they have taken, in some instances, regular photographers too, or when there is a criminal investigation or other collection of evidence. However - we all know, Adobe will certainly not implement this, or anything else, that would compete with their AI business model, unless they would make these functions very expencive, only available for big actors.

[@peinmilan]

So basically it would turn everything around : you have to proove it's real, and take everything else fake by default?!

[@FrancisJacquerye]

I hate to use buzzwords, but the blockchain could essentially provide a unique and infalsifiable identifier for each picture taken by a camera.

[@lifeexposurephotographybyt2005]

I can only imagine that this may slow down the current frame rates on photo stills. But a trade off with getting trust might still be worth it. I can see the public/private key can be used for allowing edits from our software like Lightroom and others.

[@Tarets]

But why would anyone, besides the artists themselves, care about whether something was shot and edited or ai generated? For me it is the same discussion as when digital editing came to life and purists argued it's not real photography. So what that it's not? People realizing everything could be manipulated will stop treating pictures as a proof of anything because there will anyways be a way to fake them, so why would the general consumer care about these certificates?

[@twopianomanGAB_D]

I understood this, but, for the average user ... Not sure. The average user doesn't want to do this with their PCs, MACs.

[@Scottie_McNaughty]

GO TONY!!! 🎯💯🎯💯 People have recently been thinking my photos are AI... I worked for years to perfect my style, & now I'm wondering if I should dumb them down my so people don't think they're fake 😢

[@STOVL93]

I see a lot of cynicism in the comments. I just want to say THANK YOU, Tony! This video gave me hope for how images authentication might look going forward. Your background in photography and IT put you in a unique position to explain this, and I’m glad you’re doing it!

[@coloryvr]

Because the music industry has the strongest lobby, photo, art, and video makers (along with programmers, translators...etc) should work with them to create laws and regulations. As an artist in almost all disciplines, it shocks me how each art form cries out on its own. I am not against AI, but limits should be put on its misuse as soon as possible.

[@mdturnerinoz]

This will not happen IMHO until some "important" person or group loses a sh*tton of money or someone's job/business/life.

[@jamesromanoski7292]

I really like your idea Tony. I'm a software architect with 32 years of enterprise development. Signed content with the right authentication factors (something you know, something you have, and something you are) is the right answer. It seems the real challenge is adoption. The monetary driver will be the safety provided to avoid being scammed. McAfee made a lot of money to keep people safe. I think this can do the same for digital visual content. I applaud you for taking the first steps to solving this problem.

[@scb2scb2]

He is right only apple can start it (he just missed safari as part of it) and i am sure they are working on this it seems to fit their 'you can trust us more than the other biggies' vibe. But are they willing to take the negative heat for this going at it alone at 10x the speed and force the issue. They will be called many names by many people with lots of lobby money against them.

[@DavidWilson_mfx]

I hope Apple  does do this. They are privacy-minded, and though this isn't strictly a privacy issue, it is about authenticity, and in my mind that is somewhat related. And if the social media companies followed your model of providing Check marks with verified images and an AI icon for AI-generated images, I think that other manufacturers would want to follow Apple 's lead so that photos from their cameras will also have that authenticating checkmark. People sharing images will want to show their images are verified so that their viewers will trust their images.

[@abavariannormiepleb9470]

Maybe a hot tip: Accept that 99 % of social media is worthless advertising garbage that should be ignored in general, which is also better for your mental health and only “follow” local accounts from people you actually know or from restaurants, cafés, museums, venues etc. you might visit in real life to stay up-to-date on events there.

[@daehxxiD]

I generally like the verification vs. falsification way you are proposing. I.e. assume any image is fake unless it contains the required metadata. I think editing will remain challanging no matter what (see point 2 below); but I feel like this can be solved without a "TIM"... A bit hard to formulate all of this in the comments, but I'd say we need to come up with some basic rules first: What edits are considered "safe" and which make the picture "fake" --> E.g. White balance changes, Exposure etc. is "safe" or "edited", but "Generative fill" or "Sky replacement" and/or any more advanced Photoshop technique like "replacing faces etc." makes an image "significantly altered". Maybe even some kind of threshold to how much these things are used... Like you said, using generative fill to remove a speck of dust or maybe adding a cut off finger should perhaps not count as "significantly altered", but this needs to finally be established. We were not able to do so up until now though, so I'm not sure we'll ever manage to define what a "real" photo is ;) Once this is established, editing apps could track if the ingested photo is indeed an Image with CAI signature (does this work for RAWs?) and track the changes via a hash of some sort? I think it should be doable with a client-server side authentication of that hash, instead of a "TIM". Or perhaps some other module that all hardware already has so as to not create a landfill of old camera and smartphone equipment. Couple of issues I'd raise immediatly after watching the video: 1) Right to repair nightmare: If you put that tim in there, this gives companies the ability to pretty much make anything unrepairable... Not a fan. 2) Make edits just metadata: The complexity of having image apps show modifications applied in editing apps is, in my opinion, unsolvable. Adobe changes processing standards constantly. A third party app has no way of keeping up with those changes in an economically feasible way... Let alone that would basicly mean that your images would need to be RAW files, which are way too large for instagram-type applications (that said, I think smart preview dngs are pretty small, so it COULD work)

[@tdemask]

Next up: blockchain for images…

[@Sincerelytk]

It already exists. 50% point of nfts

[@runearntzen6499]

I feel rather sure that the implementation of TIM would lead to an extra annual TIM cost. And if you dont pay, the device stops working.

[@KNZ5]

Wouldn't it be much easier if AI software were required to mark AI-generated images as AI-generated? Of course, you can take screenshots of the AI-generated images. But marking AI-generated images, will make it harder to publish them.

You have open-source AI models, that can be run on consumer hardware (like a gaming laptop). How will you force those (already released) models to comply and add the "AI" mark to generated pictures?

[@KNZ5]

You don't have to force them. It's not about eliminating AI-generated images, but about making it harder to create AI-generated images that pretend to be real, thereby reducing the prevalence of such images. Searching for the perfect solution leads to inaction, because there is no perfect solution, and you will search forever. Simple solutions that reduce fake images by, say, 80% are good enough.

[@KNZ5]

You don't have to force them. It's not about eliminating AI-generated images, but about making it harder to create AI-generated images that pretend to be real, thereby reducing the prevalence of such images. Searching for the perfect solution leads to inaction, because there is no perfect solution, and you will search forever. Simple solutions that reduce fake images by, say, 80% are good enough.

[@Solarsystem50]

I was born in the wrong time. The film years were it, man. Nobody had to worry about non of this back then.

[@scb2scb2]

You where only tricks by professionals you mean... better i guess.. maybe....

[@thierryhoornaert9950]

As a IT-er and photographer I think it is a very good idea, as fake news and fake photographs/movies are becoming a thread for society. But there are also so many creatives who 'make' beautiful composites using pictures and editing work. Think of the paintings of e.g. A. Van Dyck: the background was taken from another place, or fictional. So instead of having a 'Fake' label, I think it should be more like a 'Fictional' or 'Creative'. At the same time, (political) statements and interviews should be fact-checked and news channels should be made responsible for not mentioning it when it's fake.

[@TonyAndChelsea]

Yeah this is very valid

[@comeraczy2483]

Do you really think that when I post a picture of my dead mother on FB I want that picture to be labeled as "fictional" or "creative" because the photographer had to edit something unsightly out of the frame? Sorry but Mom isn't available to stage and authentic re-shoot. Whatever the problem that "solution" is supposed to fix, I'd be really pissed off if governments and large corporations had a word on how I am allowed to remember my loved ones.

[@davidbertrand895]

How about if you take a picture, make a print, scan the print, and submit that.

[@comeraczy2483]

Sorry, your large format pictures are all AI generated! Now, my digital picture of the Mona Lisa in my bedroom, that is certainly an authentic picture!

[@davidgrout3481]

Roughly one quarter in, it occurred to me that Tony might have purposely generated a fake video, just to show us how insidious and convincing AI can be. I do however think its real !

[@arnaudmilner238Wildbillhickock]

You known at one time people didn’t use locks. Well things change, no people changed, they stopped trusting each other.

[@briansmith5843]

I tell my kids to always use the words dad I need some chedda if they need money from me so AI won't work on me. Wait.... Damn it.

[@comeraczy2483]

15:30 "tamper-proof, like the current implementation of TPM.." is probably the funniest part. Which one of the known vulnerabilities of TPM 2.0 are you talking about? Just kidding, with TPM, no need to tamper, just wait for a compromised key, clearly labelled as "DO NOT TRUST" (not making it up, keys compromised in 2022 were still in use in July 2024, some "DO NOT TRUST" keys are probably still in use). Without that, on the vast majority of computers, a screwdriver is enough. When it isn't, it takes only a few minutes of effort to extract the corporate master password from the underpaid IT guy. My second choice is 15:50 "ideally the GPS signal would also be secure". Isn't it exactly because GPS signals are so easy to spoof that the aviation industry and the military are currently testing quantum accelerometers and gyroscopes (nope, won't fit on your phone)? The comments at 17:50 about GPS spoofing are very handwavy. Military measures against GPS spoofing aren't really a universal practical option for civilians. Anyway, are you really planning to label as fake, all pictures taken without a reliable GPS signal? GPS jamming is trivial, and there are lots of places outside of GPS range. The rest is also difficult. Keeping track of the chain of editing operations and deciding which ones are legit and which ones aren't, is plain impossible: as soon as a method is specified and implemented, tens of thousands of hackers are going to work hard on breaking it. In any case, there will be false negatives and false positives, and bad actors will use those to their advantage anyway. That said, there are a number of much more important questions: who will control the specs, infrastructure, implementation, litigation, etc.? The iphone is a monopoly almost only in the US. The rest of the world had better phones :-). Governments can't force a large number of international players to implement a good and affordable solution to a difficult problem. If they were, the outcome would certainly be ineffective, detrimental to individuals privacy, expensive, and annoying. Finally, why not do nothing about it instead? That would be a lot cheaper, a lot less riskier, and possibly more effective. Personally, I don't care at all about people disputing the authenticity of my own pictures, and I tend to avoid considering pictures as strong evidence in support of any statement. I strongly believe that the world would be a better place if more people shared these views, and from this specific point of view, TIM is a step in the wrong direction.

[@NerdSnipingBatman]

Instead of embedding authentication data in the metadata: perhaps what would be better would be to cryptographically embed it in the image itself. For several decades people have been embedding secret messages in images by ever so slightly altering the rgb values of each pixel: imperceptibly so. You could do this across the whole image to encrypt the whole image with a private key. Then when an editor or a phot viewing app wants to view the image: they need the public key which will shift the pixel values back to their correct spot. Because it will only very slightly alter the pixel values, the image could be viewed casually without needing the public key. This would have the added benefit of being able to easily detect in an inage any significantly altered parts, as that portion of the image would not have the cryptographic information. Putting TIM data in the metadata sounds too complicated and not likely to roll out. Versus encrypting photos in the pixel values themselves.

[@GemmaHentsch]

the other issue would be compression… authenticated compression/resizing would need to be available for companies like WhatsApp/instagram et al

[@genejoanen]

I'm so glad you're discussing this. This has been hard for me to keep up with we are generating all sorts of photographs for our business vote for technical purposes and for production and this has started to become a problem. I agree with other viewers about adobe. I have switched to other alternative editing software for this purpose alone. I hope we find a safe way to put our materials out that are authentic.

[@By_Rant_Or_Ruin]

Fahrenheit 451. Fascism sneaks in everywhere but mostly through good intentions and protectionism. This is the tickle that leads to the poke.

[@Sincerelytk]

This TIM solution soulds almost exact like what NFTs are except everyone tried to make fun of it and anchor on the “making money part” instead of the security of it. The blockchain does this exact thing already. But also allows you to track the sale of your photos/artwork.

[@contentm3893]

Tony - Here's the other problem. AI is so far ahead of anything that you can imagine. AI needs to be regulated and can never ever use copyrighted images, IP or licenses, period. That's the fight that every artist and video creator needs to fight. AI is already using celebrities and creating fun realistic dance videos or political propaganda parody pieces. Who are you working with for that TIM software? I know you are trying to jump on the AI train and monetize that platform. Your problem with your TIM idea is that AI generated content needs to be regulated and ALL AI CONTENT needs to be digitally marked with an encrypted blockchain. That should be on all AI companies. Meaning everything that they release has to have by law, a digital tag or blockchain. That way any platform like Meta, TikTok, etc will be able to authenticate AI content. Your software is flawed because AI will be used to take your images, vacation videos and make that content into an edited presentable piece of content. So, AI won''t be all bad because Joe Public will use AI inside of Photoshop and Premier pro and on iPhones once they capture their own real content. Does that make sense? Your TIM idea would mean everyone would need to buy your software and the next 300 copycat software systems would have to have a standard system that everyone would have to follow. Remember the VHS vs Beta wars? The same idea that you have was already thought of to stop music download piracy.

[@wolff_1]

As with a lot of things in life, too much placing the burden on people who are doing nothing wrong, there needs to be a way to place the burden on AI generators.

[@marcrjacobs]

Great concept, but getting every camera manufacturer, cell phone manufacturer and software developers to cooperate. Not likely to see that, but we do have JPGs and DNGs.

[@iVisual.sambonkowski]

we coming to a point where not only you wont need photographers, thats already the case, but you wont need a camera! sad. sad. sad

[@washingtonradio]

My concern is when round-tripping an image from the editing device to the camera is required. That seems like an awkward step to me that many will skip because they feel it's too much of a pain.

[@jamesc8517]

First you complain about Photoshop and Adobe not being good enough in manipulating images, then you want to censor all images in the universe - because your manual skill of manipulating images with photoshop falls short of AI skills? The medieval monastic book copiers (yes, they copied books by handwriting) lost their job with the invention of the printing press. Was Luther's printed bible the end of civilization? Horse carriage makers lost their job when Ford introduced the assembly line. Was that the end of private transport? Assuming every existing image is illegal unless it's de-censored by the big digital platforms is shortsighted and dangerous. Shortsighted, because photos made with chemical film for instance, which is every single photo until the nineties, don't have metadata. Dangerous, because control and censorship are the tools of totalitarians, not creatives.

[@lojeda]

We're forgetting about privacy, by adding more information to a photo we create more metadata to be linked to us

[@cous261]

I agree we already have the social media sites and everyone else harvesting our personal data and before you know it someone hacks some third party company and our data ends up on the dark web. Adding more data only narrows the data down to where those like Google Adobe Facebook Instagram and others to use the data for their profits to sell that generate more money for them to use that we get more garbage sent to us. It's a never ending process now as they wrote it into End User License Agreements which either you except, don't know or understand the changes as it's written in a loop of like a mouse on a spinning wheel going no where just as Tony pointed out with the Adobe issues of how they use our data unknowingly and we agreed to the EULA when they send them out not knowing or understanding what it is they are actually doing. An when you start digging to understand and find out here you get on the mouse wheel and run all over their website trying to figure it out and end up in the loop of endless data that never leads to a answer. It's like AI that now as we see more and more it's just getting more out of control as social media is spreading it like a wildfire and not benefitting or protecting peoples privacy as AI has already been shown to something that you can trust as it lies or puts you into that never ending loop to which a answer you never get. Honestly I don't think we can do anything to truly control this now as none of this was ever thought out in the beginning to what all the cons are to control it as like the Taylor Swift example Tony used you hire three people to where the shirts and one leaks it out they we're paid just as companies are always changing looking to cut cost and out sourcing to third parties that not all protect our data as it should be. Like oh well your money what protected by our banks system but the third party company used was the one that was hacked as they failed to protect your data and got through our system through them which looked to be real work between us and them so now your faced with who to go through to get or attempt to get your money back that's probably now gone forever. To many have access to our data and private info now days as this is just what I see and believe by doing something like this only give more accurate data to be stolen and used. Just look at how many times Google as for your location if you turn it off in Chrome that also others use this service as Google provides it as a service to them. It's my business to where I eat and what I want to buy and should not be used or sold by anyone. Just think when you by a vehicle now dealer add all their little stickers and tags on it to advertise for them and who all knows what data is being gathered through satellite as most all vehicles now have satellite antenna's for GPS services to where you go probably where your eating what stores you going to that probably is being sold to profit whomever the data is being used by. Like I said it's out of control and I'm afraid we just going to have deal with it the best we can and learn more each individually how to be more aware and teach ourselves to be more savvy if your going to live in the digital world which is becoming more and more to which the old ways are being done away with. For just as Tony and many have also pointed out the more AI grows the less photographers, videographers as many other jobs are going to be less needed for AI will do it all as were just putting ourselves out of work becoming more dependent on machines and less of the human intervention. It more like a Super Covid that their will be not cure unless man starts taking back the work and depending less on machines though they can be used to assist as they should be and nothing more. Its a sad world I see coming more and more everyday as we become more machine dependent and letting AI do things man should be doing it's like the old days when we watched Dick Tracy, Star Trek, and Inspector Gadget as we believed the things we seen like Dick Tracy talking through his watch to be unreal yet it's here along with things like monitoring our medical health. All things we grew up watching we never believe could be real but thought was cool and now cool is here draining our funds because it's the latest and greatest we have to have but do we really I ask myself now and finding more and more not really but the old things are dying because companies are changing to make us use the latest things and not maintaining the old that still will work just needs to be updated as the new stuff is everyday and we end up filling the landfills with more of it to which we also are destroying or world as we do. Just look at plastic which they said would be recycled yet now we are finding it just being spread across the world in our oceans and landfills. And now they want more money to help with recycling another mouse in the wheel game.

[@STOVL93]

What privacy concerns do you see other than GPS metadata, which Tony said should be off on default except in journalistic circumstances?

[@ericsart6474]

With any gate keeping system, it with inevitably keep out something that is legit. This is the problem with gatekeeping. It also will rely on human interpretation of what the criteria should be.

[@aviationmedia2425]

Tony, this leads to big brother systems and governments will abuse it.

[@TechLeatherCraft]

I was thinking the same... Who is going to implement, scan, (and collect) all this "Tim" data... Adobe, Facebook/Instagram, Google/Apple... Slight problem with that.

[@STOVL93]

You’re already uploading all that stuff every time you post an image (unless you scrub the metadata, which almost nobody does). Tony compared it to SSL. Being able to verify I’m connected to my bank’s website before I type in my password is crucial-and it’s got nothing to do with government overreach.

[@Hodenkat]

You mean capitalists will abuse it. Only those who can figure out a way to make profit from it will care.

[@comeraczy2483]

We don't even need big brother to get abject outcomes: imagine posting your favorite picture of a lost loved one, and see it presented with a watermark "Fake: AI generated" because the photographer had to edit something unsightly out of the picture.

[@ChristopherJohnDotCom]

We don’t need more surveillance. TIM sounds like a bad idea.

[@JasonMacen]

The server side verification processing would be 100x that of ssl.

[@tubularificationed]

Luckily, the camera industry has shown to be capable to talk to each other 🙂 There is already this JEITA organisation which created the Exif metadata standard for cameras (actually for any imaging devices, including e.g. scanners). Maybe they should create a TIM task force?

[@tonywhite3124]

South Australian Government is wanting the community to comment on deepfake and AI content... I have used this video as exhibit A (by sending the commission a link to this video) hopefully other government organisations do similar.. Keep up the good work T+C

[@pxperimenter]

All you need to defeat a TIM chip is to create a fake AI video or image, play it back on a display, and record it with a TIM-chipped camera - voila, you'd have a TIM-authenticated fake, and the system would do more harm than good.

[@unstanic]

Plot twist, TIM will utilize AI to detect screen recording

[@TonyAndChelsea]

I address this

[@ANSWERTHECALLOFJESUSCHRIST]

Ah, remember the simple days? "That image is Photoshopped" was as far as we had to worry. 😅 EDIT: By the way, what is it with Tony and his Hugh Brownstone-esque speech cadence in this video? 🤔

[@rpmorrisjr]

Haha. I’m old enough to remember photos being “airbrushed”. I’m not even that old.

[@ScottJWaldron]

7:50 Embedding image data from multiple camera modules is a smart idea. 👍

[@paul-travels]

WRT CRLs. If the Private Key somehow gets leaked or the CRL is incorrectly updated to include a particular camera, what happens with that camera? I assume that it will be able to still take photos but now none of the images are trusted. Seems like it would impact the value of that camera, no?

[@phillipmaguire4671]

It's all becoming too much Tony... People can no longer trust social media or know if the images they see are real. Also, I am fast becoming "screened out". Sick and tired of living a life via a screen. I just now want out from social media too. I only want to connect with people face to face, and forget about looking at an image on a screen.

[@gamebuster800]

8:57 so this "TIM" would be responsible for processing the image? That would need quite some processing power for a security oriented chip. I like suggestion 2 much more: Adding edits to the metadata. However, I can imagine the implementation would be a nightmare as the result can vary based on subtle differences. Imagine trying carefully to edit an image on your device, and it ends up rendered differently because the Facebook app uses a different algo to process the edits. It also means you share the full size RAW file with anyone, including the edits. Many photographers seem to rely on reduced resolution and detail of a JPEG to protect their source files. RAW files are also much much bigger than cropped, downscaled JPEGs. That's a lot of hassle for something that can be faked anyway. One "broken" TIM and poof, you can generate fake images anyway... So you keep a list of "trusted devices". What if one device proves to be vulnerable? Are you going to invalidate all images taken by these devices? And all of this completely ignores video. This "TIM" would scale poorly with video. You can't just embed full "RAW" videos with edits as metadata, it's just too much data. A TIM processing videos requires a pretty powerful chip.

[@looseshoulderssumali7311]

So...put TIM inside photoshop, lightroom, photo mechanic, capture one, evoto, etc?