Someone Is Pretending To be Me Right Now | by CredibleCrypto
Literally. As I write this, there are hundreds of accounts on X using my name, logo, bio, and banner, messaging my followers about “special opportunities” to trick them into sending “me” money in one way or another.
Since August 24th 2026, when I began working with Umanitek and their “Guardian agent” service, over 500 fake CredibleCrypto accounts have been flagged and taken down. That’s 500+ impersonating accounts in literally two weeks. Every one of those takedowns carries an evidence file published to the OriginTrail Decentralized Knowledge Graph (DKG)- the public, verifiable infrastructure that all of Umanitek’s products run on.
For years, my only defense was my selfless followers who took time out of their day to tag me under fake accounts so I could manually block and report them. Sometimes these followers would simply report these accounts themselves. Either way, the only way to try to combat this problem was for me (or my followers) to report these impersonators one at a time. Meanwhile, the platforms took days to act on these reports, while three new fakes spawned for every one removed. I was simply fighting a losing battle, despite all the help I had from my followers.
Fast forward to March of this year, and I came across a post from Umanitek saying they were looking for beta testers to try their new “Guardian Agent” product, which they had been developing for some time. After literally years of dealing with the endless impersonation problem, I jumped at the opportunity. Shortly after, I was actively engaged with the team- trying out the new Guardian Agent product, sharing feedback on the internal dashboard and UI, and figuring out the best way to implement a solution to this seemingly never-ending problem. In early August, nearly 6 months later- and thanks to all the progress I made with the team while beta testing- I was offered the opportunity to become the first official Umanitek “product ambassador.” This opportunity has allowed me to continue to get deeper insight into the inner workings of Umanitek products during development while at the same time allowing me to work closely with a team that is solving real-world and relevant problems, while actively building on the OriginTrail Decentralized Knowledge Graph (a network I’ve personally been a vocal supporter of since 2021). I gladly accepted this opportunity, and so I’m starting this article series today to share what I have learned about Umanitek and OriginTrail, and how they are working together to fight the rapid proliferation of AI abuse we are seeing on the internet today.
This first article covers the three tools that Umanitek has developed in their current product suite. Next month, we will go under the hood of that infrastructure to examine how the OriginTrail DKG actually works and what role $TRAC plays in it. Finally, in month three, we will tie everything together so you can better understand the implications and potential of this new yet vital tech.
Disclosure: This series aims to explain what this technology does and how it works for educational purposes, from both technical and non-technical perspectives, as simply as possible. It is not financial advice. As always, do your own research.
Something has fundamentally changed about how identity theft, scams, and fake content work online. Today, a company called Umanitek is stepping forward to fight it by building what I can only describe as an “immune system” for the internet- a set of tools created as the first line of defense against impersonation, misinformation, and outright scamming on the web. Part of being a truly effective and unbiased solution to this problem is having a decentralized, publicly auditable memory layer to record each action/event that is taken by these tools to ensure that no single entity can own, influence, or compromise the system- and that is where OriginTrail comes in. Umanitek's product suite runs on the OriginTrail DKG, a public, decentralized knowledge graph secured by a token called $TRAC. Every product you are about to read about records its evidence on this knowledge graph.
The Problem Has Grown Too Big for Human Solutions
Here is the hard truth of the AI era:
Creating convincing fake content used to require much more effort. Cloning a profile, sure, that was always easy- but deepfaked videos, cloned voices, AI-written DMs that sound exactly like the person they’re impersonating? That took a number of complex skills. Now, it takes a single prompt.
The cost of producing damaging content has collapsed to nearly zero, while the cost of fighting it hasn’t budged. Before Umanitek, every takedown still required a human to notice, report, and submit a takedown request for infringing or “fake” content, hoping the hosting platform would recognize the abuse, accept the complaint as valid, and ultimately remove it. This entire process had to be done manually, but because of how easily this fake content could be created en masse, this was like shooting squirt guns at a forest fire.
And it’s not just impersonation scams. The same cost collapse applies to deepfakes of public figures, manipulated media in news cycles, stolen intellectual property, and the darkest categories of damaging content imaginable. The tools that made creating this type of content effortless also made abuse effortless.
It has now become abundantly clear- you cannot fight AI-scale problems with human-scale solutions. The defense has to operate at the same speed and scale as the attackers. And because of the potential for abuse by biased parties, the defensive network also has to be independently verifiable by people who have no reason to trust each other- which is why Umanitek built on public infrastructure via the OriginTrail DKG instead of a private database of its own. That’s the premise behind everything Umanitek builds in pursuit of their stated mission- to strike through damaging content, IP infringement, and fake news, and make internet safety a standard in the age of AI.
Right now, their product suite consists of three tools. All three products share a single design decision, and that's why this series is called Umanitek × OriginTrail. None of them keep their evidence exclusively in a centralized, siloed company database. Rather, they write it to the OriginTrail Decentralized Knowledge Graph- a public, permissionless network where records are anchored, timestamped, and impossible to edit or delete later on.
This turns a simple claim into valid, effective proof. Umanitek builds the agents, but the DKG records every action the agents take and lets you validate their work without taking anyone’s word for it, including Umanitek’s. Meanwhile, $TRAC is the token that pays for and secures that entire network. (Keep this in the back of your mind as we go through each of the three tools in the next section, as they're all natively built on this infrastructure.)
Guardian: Your Personal AI Agent Watchdog
Guardian is Umanitek’s flagship product- a personal AI agent for identity protection and the product responsible for my 500+ impersonator takedown tally. Think of it as a security guard for your digital identity, except this guard watches hundreds of millions of accounts across sites like X, YouTube, Facebook, and TikTok simultaneously, around the clock. It’s looking for anyone using your face, your name, your bio, or your image- the raw materials of impersonation.
Guardian tracks three kinds of risk:
Fake accounts and identity theft. The classic scam-account problem that I was dealing with myself. Fake accounts using your image or likeness to impersonate you (usually with malicious intent). Guardian detects profiles built from your identity, scores them by “danger” level, and moves to remove them.
AI misrepresentation. A newer, stranger risk: what do AI models say about you? When someone asks a chatbot about your name or brand, false or misrepresented information becomes a reputational attack surface that most companies/brands haven’t even considered yet. Guardian searches for potentially misleading information and brings it to your attention so you can decide how to move forward and address it.
Content and IP infringement. Your videos, thumbnails, or other work- reposted or monetized by someone else. Guardian identifies this misappropriated content and flags it for removal by sending official takedown notices to the relevant hosting platform on your behalf.
In each of these instances, Guardian doesn’t simply track and remove the damaging content; it also maintains all the evidence on a permanent, immutable knowledge graph.
When Guardian finds a fake, it doesn’t just flag it- it builds a verifiable evidence pack documenting when the account appeared, where it operates, and what it’s doing. That evidence pack is published to the OriginTrail DKG, where it can’t be edited or deleted after the fact and where a platform, a lawyer, or a complete stranger can verify it without asking Umanitek for permission (more on how that works in Article 2).
Why does this matter? Because a takedown request is only viable if clear, unbiased evidence backs it up. A screenshot and an angry report are easy to ignore. A complete, timestamped, independently verifiable case file is not- whether it lands on a platform’s trust-and-safety desk or a lawyer’s. Guardian turns what would normally be a “simple” takedown report into a full-fledged legal request- something that content moderation teams are likely to take a LOT more seriously.
So when I say over 500 fake Credible accounts have been taken down, I also mean that 500+ evidence reports have been created, documented, and saved on the OriginTrail DKG to back up each and every takedown request filed. This is likely the main reason these takedown requests have been so successful at removing fraudulent accounts.
Guardian has probably been my most-used and most effective Umanitek tool that I’ve used so far, primarily because of the impersonation issues I deal with as a decently sized content creator on the internet, but the next tool in Umanitek’s arsenal is one that I believe every single internet user today will find to be incredibly relevant and helpful.
This second tool is called “TrueSeal”.
Let’s dive right into it.
TrueSeal: The “is this real?” Machine
The second product in Umanitek’s product suite, TrueSeal, launched in June 2026, and it attacks the problem from the opposite direction.
While Guardian protects your personal identity/brand, TrueSeal focuses on all the other content that can be found on the internet- namely, whether or not you can trust what you see.
We’ve all had the moment: a video crosses your feed- a public figure saying something explosive, a “leaked” clip, a too-perfect image or video- and your gut asks, is this real? A few years ago, your eyes could usually help you arrive at the correct answer. Today, honestly? They can’t. Deepfakes and manipulated media have blurred the line between fiction and reality.
TrueSeal is built for exactly these moments. Feed it a suspicious post, and it performs verifiable image and video analysis, then turns what it finds into clear, inspectable evidence that anyone can act on. Not a black-box “87% likely fake” score you have to take on faith, but an auditable result- one that’s delivered on public, tamper-proof infrastructure (the OriginTrail DKG, where anyone can check it), rather than Umanitek’s own servers.
That distinction is key. Plenty of tools claim to detect deepfakes. TrueSeal’s difference is that its analysis produces evidence you can inspect and verify independently- every verdict comes with a receipt. In a world where you can no longer trust what you see with your own eyes, an auditable tool like TrueSeal becomes a near necessity.
For creators and individuals in the public sphere, this means being able to prove that a “smear” video is fake. For journalists, it means being able to verify a source clip as genuine and authentic before running with it. For everyone else, it’s the ability to check an image or video on your social media feed before you share something that was deliberately engineered to fool you (and potentially make a fool out of you).
TrueSeal is probably the most important tool in Umanitek’s suite for both the average/casual user and high-profile persons alike.
The third, and final tool in Umanitek's product suite is designed to make the network effects of the internet and AI agents work for you, rather than against you.
That tool is the Guardian Blackbox.
Guardian Blackbox: Safety as a Network Effect
The Guardian Blackbox is an open-source, permissionless version of the Guardian Agent that anyone can run, leveraging network effects to protect everyone who runs it.
Every individual Blackbox agent that identifies a threat- a scam account, malicious content, a fraud pattern, etc.- syncs that intelligence, anonymized, to the OriginTrail DKG. Every other Blackbox on the network then syncs from the same graph, creating a decentralized network of guardian agents working together to protect the entire Guardian Blackbox user base.
In other words, when one node learns, every node learns. A scammer flagged by one community is instantly known across all communities.
Compare that to how internet safety works today: Each platform keeps its own private blacklist. Privacy law, competition, and liability mean they can’t (or won’t) share threat data with each other. So a scammer banned on one platform can simply walk next door and start fresh on another.
The Blackbox changes this dynamic. Because threat intelligence lives on neutral, public infrastructure- owned by no company and editable by no individual- everyone can contribute to it, and everyone can verify against it, without anyone handing over their private data. In essence, every person running a Blackbox Guardian automatically makes every other user within the network safer.
As a crypto native, I find the decentralized, permissionless, and tamper-proof structure behind Blackbox Guardian familiar and recognizable, and that’s no coincidence. As mentioned earlier, every Umanitek tool has been built around the “don’t trust, verify” ethos that underpins the broader crypto movement.
“Don’t trust, verify”
Now that we’ve covered all three of Umanitek's products, let’s take a step back and look at the recurring theme of each product side by side:
- Guardian doesn’t just identify and claim an account is fake- it produces a verifiable evidence pack with which to validate these claims, rather than expecting a content moderation team to trust them blindly.
- TrueSeal doesn’t just score a video with some ambiguous claim as to its authenticity - it delivers an auditable, inspectable result, so that one can verify exactly why it has come to this conclusion.
- The Guardian Blackbox network doesn’t just share threat intel- it shares this intel on the OriginTrail DKG so that every instance is recorded, auditable, and verifiable on a public knowledge graph that no one can rewrite or manipulate.
One principle runs through the entire suite: in the age of AI, “trust us” is no longer good enough. Every claim these products make comes with receipts that anyone- you, a platform, or even a court of law- can independently check.
Before reading this article today, you might have asked: what kind of infrastructure makes that even possible? What system lets competing platforms, independent researchers, and random individuals all verify the same evidence without trusting each other or any central company and without exposing sensitive data?
Now you know this infrastructure is the OriginTrail Decentralized Knowledge Graph, which is powered by a token called $TRAC.
The $TRAC Token
Remember that every evidence pack a Guardian agent files, every TrueSeal verdict, and every threat signal a Blackbox shares gets published to the DKG- and writing to the DKG is paid for exclusively in $TRAC. This has been the case at the protocol level since the token's inception. However, exciting developments are on the horizon: OriginTrail’s developers have now validated, across a series of prototypes, that DKG queries and inference can be metered, served, and settled in $TRAC as well- node-to-node, in both directions. This introduces a completely new use case for the $TRAC token. For tools like the ones built by Umanitek, reading is the busier half of the equation- a platform checking a Guardian case file, every Blackbox syncing what every other Blackbox has learned, an AI agent querying the DKG before it answers a question about you- all of these actions will soon be able to be executed and paid for with $TRAC. $TRAC’s first job was as a payment mechanism used for publishing/writing knowledge to the DKG, but it’s now getting a second one as a payment mechanism for inference queries and requests on what I strongly believe is one of the most important pieces of infrastructure in the entire AI era- the OriginTrail DKG.
That is the topic we will dive into in much more depth for Article #2 next month, so stay tuned!
Note: If you or someone you know is actively dealing with impersonation online and would like some help in finding an effective solution to this problem, feel free to reach out and I'll be able to direct you to someone from the Umanitek team to work on a custom solution.