Buscar
Buscar
July 4, 2025
Combating AI Deepfakes in Web3 Identity With Blockchain
Compartir ...
Por: Piero Matias
The rapid rise of AI-generated deepfakes is our new reality in 2025. Today, you can generate ultrarealistic fake videos, images, or audio almost without any technical background and with payne a couple of dollars for AI service subscription. This easiness is fueling a new kind of identity fraud when hackers use AI generated media to trick their victims. Deepfakes use machine learning to imitate a real person’s look or voice with amazing accuracy.
This approach is starting to be used very widely for scams like impersonating relatives or colleagues to steal money or corporate data. If you ask your friends and colleagues, it will turn out that over 90% of people are now worried about the spread of deepfakes, and experts warn that this synthetic media is one of the biggest threats to the digital world today.
For Web3 communities the deepfake is even a bigger problem. How do you trust that someone on the blockchain or in a DAO is a genuine person and not an AI-generated avatar? Using sophisticated bots and deepfakes has granted hackers with almost unlimited possibilities in fraud or Sybil attacks when one person is controlling many fake identities.
The good news is that the same technology powering Web3 is now being used to fight back against deepfakes. And this technology is Blockchain! This approach and new security practices are starting to be implemented in the KYC process for banks, crypto projects, casinos like 777bet and many other businesses where you need to know your clients before you can open accounts for them.
Anyone can use GANs or other generative networks to create deepfakes based on real data to produce fake content that looks and sounds hell real. Any specific AI may study hundreds of photos of a person to fabricate a video of them speaking. Just look at this Elon Musk video on YouTube that was produced 2 years ago and took 70 hours of rendering and compare it to this video generated by Google’s VEO3 in a couple of minutes.
VEO3 and all the rest of similar services will double the quality by each 6 months. So in 6 to 18 months from now, fake media will become hardly distinguished from reality. We’ve already seen deepfake audio used to mimic a couple of CEO’s voices and now media reported a couple that was duped by an AI-generated “relative” on the phone, paying out thousands in a scam.
These incidents highlight a chilling truth: you cannot believe what you see or hear online without revalidation in the deepfake era.
To counter this threat, businesses are turning to blockchain and decentralized identity (DI) technologies as a solution against deepfakes. Blockchains are tamper-proof ledgers – they excel at verification and secure record-keeping. A recent Forbes report noted that using blockchain for digital IDs provides verified proof of identity and “adds trust and safety to interactions online.” Decentralized identity tools from companies like Polygon (Polygon ID), Nuggets, and Unstoppable Domains are already offering verified credentials for users’ digital personas.
Such blockchain-verified profiles act as a kind of “proof of humanity” in Web3, allowing users to store trusted identifiers and credentials under their control. Even if a deepfake perfectly mimics someone’s face or voice, it won’t possess that person’s cryptographic identity proofs. Below, we explore several key blockchain-based approaches – in accessible terms – that are being used to prevent or detect deepfakes in the context of Web3 identity.
It’s worth highlighting a few of the projects bringing these ideas to life. We’ve mentioned Worldcoin (scanning irises and using ZK proofs) and Humanity Protocol (scanning palms and issuing private “Proof of Humanity” credentials). Both aim to create Sybil-resistant digital identity networks.
Biometrics are another tool being secured with blockchain tech to outsmart deepfakes. The idea is to use a real human’s biometrics to establish a one-of-a-kind digital identity, but do it in a privacy-preserving way by storing only hashed or encoded data. For example, Worldcoin’s sign-up process involves an Orb device scanning your iris (the colored part of your eye). This scan captures the unique pattern in your iris and converts it into an IrisCode – basically a numeric representation of that pattern. Importantly, instead of saving a picture of your eye, the system stores a cryptographic hash of the IrisCode. A hash is a one-way fingerprint; Worldcoin’s hash can’t be reverse-engineered to reveal your actual iris image, but it can be used to check if the same iris has already been seen.
In Worldcoin’s case, over 2 million people have already been iris-verified for a World ID, and Humanity Protocol just hit 1 million verified users in testnet, reflecting huge demand for these solutions. Another example is BrightID, which doesn’t use biometrics but builds a unique-identity system through social graph analysis: users join via video calls and mutual connections in a web of trust, making it hard for one person to pretend to be many. For content authenticity, tech giants and startups alike are joining the effort. Adobe’s CAI is working on open standards for media provenance, and projects like Truepic and Scarlet use blockchain to timestamp and seal images at capture.
On the decentralized identity front, there’s Polygon ID that provides a toolkit to create zero-knowledge identity proofs for dApps. For example, you can prove your own verified credential (KYC token, age attestation, etc.) without revealing it.
Civic (Identity.com) offers a blockchain-based ID verification platform that can perform liveness checks and issue reusable verified credentials to your wallet, which you can later present to prove “I’m verified by Civic” rather than showing an ID document. And enterprise-focused startups like Nuggets combine biometric user authentication with decentralized ID storage, so businesses can confirm a user’s identity continuously without ever holding the user’s raw biometric data.
This approach is starting to be used very widely for scams like impersonating relatives or colleagues to steal money or corporate data. If you ask your friends and colleagues, it will turn out that over 90% of people are now worried about the spread of deepfakes, and experts warn that this synthetic media is one of the biggest threats to the digital world today.
For Web3 communities the deepfake is even a bigger problem. How do you trust that someone on the blockchain or in a DAO is a genuine person and not an AI-generated avatar? Using sophisticated bots and deepfakes has granted hackers with almost unlimited possibilities in fraud or Sybil attacks when one person is controlling many fake identities.
The good news is that the same technology powering Web3 is now being used to fight back against deepfakes. And this technology is Blockchain! This approach and new security practices are starting to be implemented in the KYC process for banks, crypto projects, casinos like 777bet and many other businesses where you need to know your clients before you can open accounts for them.
Why Are Deepfakes So Dangerous?
Anyone can use GANs or other generative networks to create deepfakes based on real data to produce fake content that looks and sounds hell real. Any specific AI may study hundreds of photos of a person to fabricate a video of them speaking. Just look at this Elon Musk video on YouTube that was produced 2 years ago and took 70 hours of rendering and compare it to this video generated by Google’s VEO3 in a couple of minutes.
VEO3 and all the rest of similar services will double the quality by each 6 months. So in 6 to 18 months from now, fake media will become hardly distinguished from reality. We’ve already seen deepfake audio used to mimic a couple of CEO’s voices and now media reported a couple that was duped by an AI-generated “relative” on the phone, paying out thousands in a scam.
These incidents highlight a chilling truth: you cannot believe what you see or hear online without revalidation in the deepfake era.
How to Protect Yourself?
To counter this threat, businesses are turning to blockchain and decentralized identity (DI) technologies as a solution against deepfakes. Blockchains are tamper-proof ledgers – they excel at verification and secure record-keeping. A recent Forbes report noted that using blockchain for digital IDs provides verified proof of identity and “adds trust and safety to interactions online.” Decentralized identity tools from companies like Polygon (Polygon ID), Nuggets, and Unstoppable Domains are already offering verified credentials for users’ digital personas.
Such blockchain-verified profiles act as a kind of “proof of humanity” in Web3, allowing users to store trusted identifiers and credentials under their control. Even if a deepfake perfectly mimics someone’s face or voice, it won’t possess that person’s cryptographic identity proofs. Below, we explore several key blockchain-based approaches – in accessible terms – that are being used to prevent or detect deepfakes in the context of Web3 identity.
Real-World Examples and Innovations:
It’s worth highlighting a few of the projects bringing these ideas to life. We’ve mentioned Worldcoin (scanning irises and using ZK proofs) and Humanity Protocol (scanning palms and issuing private “Proof of Humanity” credentials). Both aim to create Sybil-resistant digital identity networks.
Biometrics are another tool being secured with blockchain tech to outsmart deepfakes. The idea is to use a real human’s biometrics to establish a one-of-a-kind digital identity, but do it in a privacy-preserving way by storing only hashed or encoded data. For example, Worldcoin’s sign-up process involves an Orb device scanning your iris (the colored part of your eye). This scan captures the unique pattern in your iris and converts it into an IrisCode – basically a numeric representation of that pattern. Importantly, instead of saving a picture of your eye, the system stores a cryptographic hash of the IrisCode. A hash is a one-way fingerprint; Worldcoin’s hash can’t be reverse-engineered to reveal your actual iris image, but it can be used to check if the same iris has already been seen.
In Worldcoin’s case, over 2 million people have already been iris-verified for a World ID, and Humanity Protocol just hit 1 million verified users in testnet, reflecting huge demand for these solutions. Another example is BrightID, which doesn’t use biometrics but builds a unique-identity system through social graph analysis: users join via video calls and mutual connections in a web of trust, making it hard for one person to pretend to be many. For content authenticity, tech giants and startups alike are joining the effort. Adobe’s CAI is working on open standards for media provenance, and projects like Truepic and Scarlet use blockchain to timestamp and seal images at capture.
On the decentralized identity front, there’s Polygon ID that provides a toolkit to create zero-knowledge identity proofs for dApps. For example, you can prove your own verified credential (KYC token, age attestation, etc.) without revealing it.
Civic (Identity.com) offers a blockchain-based ID verification platform that can perform liveness checks and issue reusable verified credentials to your wallet, which you can later present to prove “I’m verified by Civic” rather than showing an ID document. And enterprise-focused startups like Nuggets combine biometric user authentication with decentralized ID storage, so businesses can confirm a user’s identity continuously without ever holding the user’s raw biometric data.