I know this has probably come up before, but in light of the recent prototype scandal, the Akabane extra copies surfacing, rapid advances in AI, and the sheer amount of money tied up in holy grail cards, it feels almost inevitable that someone will eventually develop a foolproof way to counterfeit—even cards with the most difficult holos to replicate. What do you think? Could this become an existential risk to the hobby? And if so, will PSA and the other legacy graders ever find a way to truly future-proof against it, or is this a problem that only more AI-native companies like TAG will be able to tackle?
All grading companies have graded fake cards, fake stamps, fake autographs, etc over the years. Things like autographs are scary, there is likely a machine out there that can perfectly replicate any autograph with correct pressures applied at the correct places after teaching it from a few real examples.
5 or 50 years from now, there is likely an ability to reproduce any card without detection from human or computer interaction. At that point, what is the real cardboard worth?
3 Likes
I think its another reason why we need to move away from human graders – In theory, an AI trained on millions of cards could spot invisible tells in print layers or holo diffraction that a human could never see.
Maybe.
Counterfeiters tend to follow the path of least resistance to quick money though. There are much easier things to counterfeit with much less scrutiny on them to keep them occupied for the foreseeable future.
Time will tell
3 Likes
I don’t want to be harsh, but I don’t think you understand what AI is and how it works. AI is not going to be able to magically print exact replica cards, nor will it develop simultaneous clairvoyance and x-ray vision to detect fakes.
We already have advanced TCG authentication methods like CGC’s (when they use it). These automated/tech tools are assisted by humans to determine whether a card is real or not, and they’re very effective. I’m sure PSA has similar methods, although they don’t do the public information posts that CGC does. I would think a “take a picture of the front and back” company like TAG is far behind PSA and CGC’s authentication practices, as much as they try to promote themselves with “AI” as the tagline for everything.
Us Pokemon collectors are often laser-focused on our own hobby, but as @Mr.Garrison points out, we can imagine what could happen if it did become possible to fake cards convincingly. Why would Pokemon cards be the target when sports cards are orders of magnitude more expensive and less challenging to fake? A Mantle/Gretzky rookie has no holo and is just ink on cardstock. If you have the ability to make a 1:1 copy of a Pokemon card, you can probably just cut out the middleman entirely and counterfeit currency.
14 Likes
Replicating the elemental composition of the original materials down to the atomic distribution of heavy metals in the materials used would be near-impossible. This can be tested using X-ray fluorescence spectroscopy. Even if that was overcome, would these counterfeits react similarly to infrared and ultraviolet light? Are the card dimensions and thickness accurate? Would the offset printing plates be correctly aligned (i.e., screen angles)? etc. etc. etc.
The point I am trying to make is that it would be near-impossible to perfectly replicate vintage cards that used materials that are no longer available. The existential risk comes from ex-employees selling authentic sheets, original graphics files, etc. to reproduce the cards with authentic materials.
Documentation and detection needs to improve considerably to ensure that the cards being graded were made using authentic materials from the correct time period. Increasingly, provenance should come into picture for ultra high-end cards if the end-buyer cares whether it was an Akabane closet copy or not.
AI is fully irrelevant to this conversation.
5 Likes
If technology advances to the point where it’s possible + cost-effective to produce counterfeits that are literally indistinguishable from real cards, we have much bigger concerns than Pokemon cards. Like, what about the value of money in the first place? E.g., perfect paper currency counterfeits could lead to the total meltdown of the global financial system.
5 Likes
I have a background in machine learning so let me add a different perspective here. The issue isn’t that AI is going to magically grow X-ray vision or clairvoyance. The real risk is that generative models and multimodal systems are getting very good at replicating complex textures, holographics, and print patterns. What used to take expert counterfeiters months of trial and error can be accelerated by AI-assisted imaging and simulation. It’s not magic, it’s automation closing the gap between real and fake much faster than before.
I agree that PSA and CGC have solid methods today, but authentication is always an arms race. The same tools graders can use to spot fakes can also be used by counterfeiters to remove those tells. That’s the dynamic in every field from cybersecurity to art authentication to deepfakes. Saying “current methods are effective” is true for now, but the ground shifts quickly.
And about TAG — it’s not really about just snapping a picture. The idea is to build a massive reference library of microscopic surface features, ink distributions, and print artifacts across millions of cards. At scale, that kind of dataset could give AI a huge advantage over human graders who still rely mostly on heuristics and experience.
Finally, Pokémon would still be a prime target even if sports cards are worth more. Pokémon is more liquid, more culturally iconic, and comes with fewer legal risks compared to faking licensed league products. Counterfeiters don’t only chase the highest dollar amount, they go after markets that are both profitable and scalable. Pokémon fits that profile perfectly.
So I don’t think it’s fair to dismiss AI as irrelevant here. The risk isn’t science fiction, it’s the speed of the arms race getting faster, and the hobby has to be realistic about that.
2 Likes
I get where you’re coming from, and you’re right that replicating vintage materials down to the elemental level is extremely difficult. But the challenge isn’t necessarily about getting a perfect replication of 1990s print stock. From a technical standpoint, AI-assisted counterfeiting doesn’t need to be flawless at the atomic level—it just needs to get “good enough” to fool the majority of graders and collectors most of the time. That’s the real risk.
AI is already used in materials science to model and simulate composites, inks, and coatings. Pair that with access to modern printing equipment, and the gap between “close enough” and “indistinguishable” starts shrinking faster than we think. I agree that ex-employees selling old sheets or plates would be a worst-case scenario, but you don’t even need that for the arms race to accelerate. It’s the combination of advanced reproduction tech and the ability to iterate thousands of times digitally before ever touching a press that changes the landscape.
In the end, provenance and chain-of-custody might matter just as much as physical authentication. But I don’t think we should assume “near impossible” is a permanent state—AI has a way of turning “impossible” problems into engineering challenges over time.
For both counterfeits and use of ai, only time will tell but counterfeits certainly are improving at an alarming pace. I believe this is more of an issue for modern as vintage cards have been known to be faked and therefore are subject to more scrutiny.
2 Likes
With all due respect, the biggest impact from AI on this hobby is that people such as yourself are using it to generate generic replies like this lol.
AI is going to be the death of forums like this. Everything becomes ultra-generic, approximately correct (but rarely precisely correct), and in high school essay format 
.
1 Like
I think dyl got it right.
Fakes will keep getting better but so does our ability to detect them. Counterfeiters only have to make the cards “good enough” to convince one person.
I have more worries about unauthorized printings of modern cards from the source files by rogue employees than an external counterfeit (basically what happened with the prototypes too)
10 Likes
AI is not going to magically improve printing or recreate the offset printers.
I don’t think AI grading is the best solution, well trained humans using multiple methods of authenticating can spot the vast majority of counterfeits very rapidly.
I am waiting for one of the companies to introduce XRF to all cards above a certain dollar value. It is the easiest and most definitive way to detect anomalous cards, ink, etc.
5 Likes