The popularity of non-fungible tokens has soared, and as with everything else that becomes popular in the crypto world, fraud is on the rise too. As a result, many people see NFTs as a potentially risky investment due to the potential for being scammed. Luckily, the industry is fighting back with the introduction of machine learning-based technologies that can be a game changer for NFT verification.
NFT fraudsters use a number of techniques in order to cheat buyers, brands, artists and collectors. The most common of these include phishing scams, pumps-and-dumps, fake persona scams and NFT marketplaces, and blatantly counterfeit NFTs.
NFT fraud is thus a costly business, but the good news is that multiple projects are working to solve the problem. They’re doing this through a combination of machine learning and powerful incentives to create a more secure environment for NFT creators and collectors to collaborate.
Most experts on the NFT scene will stress how it’s important to “do your own research” when deciding whether or not to invest in a new token. However, the reality is that a large number of people who’re interested in buying NFTs simply don’t have any idea how to research a token’s authenticity, so they inevitably have to make an uninformed decision. New solutions that employ machine learning have enabled the creation of algorithms that are trained using historical NFT transaction data, so that they can spot fraudulent tokens and scams.
The use of machine learning to identify fraud in other industries, such as financial services, insurance and gaming has been well documented, and the same tried-and-tested techniques can also be applied to blockchain. Machine learning algorithms provide a more effective and streamlined way to detect fraud thanks to their ability to handle large volumes of data. In this way, they’re able to detect suspicious trading patterns far faster than any human can.
Another benefit of machine learning models is that they’re more accurate. Because they can process enormous volumes of training data and constantly train themselves, they become increasingly accurate over time. They’re more flexible too, as they’re able to perceive datasets in ways that humans would never be able to.
Perhaps the main advantage of machine learning-based fraud detection is its cost and scalability. Such systems are extremely scalable and they do not age. Rather, they become better at predicting fraud over time as the algorithms ingest more data.
A number of ML-based NFT fraud detection tools have emerged in recent years that share the goal of helping creators and collectors alike to defeat scammers. One of the most popular such protocols to emerge is WatchDog, which uses algorithms to identify duplicate NFTs and trademark infringements. It leverages powerful computer vision models that enable it to spot duplicates, even when an image or text has been altered. Whenever it spots fraudulent activity, it can notify the intellectual property rights holder and suggest ways to protect that IP.
Watchdog’s tools include an AI engine that monitors blockchains such as Ethereum in real-time to detect suspicious activity, plus a notification service that works with Discord, Twitter, Telegram and email to alert IP rights holders to potential fraud. It can also provide a detailed report on NFT scams.
Doppel meanwhile is a cross-chain NFT monitoring platform that’s designed to seek out counterfeit tokens. It works by indexing NFT data from multiple blockchains, including Ethereum, Flow and Solana. It then scans these chains for new NFT activity, vetting it against its vast, existing datasets of NFTs to identify when counterfeits are created.
Finally, PixelPlex recently launched a tool called CheckNFT.io, which can be used to analyze NFT collectibles, detect fraudulent activities and tokens, and minimize risks.
With machine learning-based NFT fraud detection tools now widely available, protocols like Wakweli promise to make a real difference in the battle against NFT fraudsters. Wakweli, which is named after the Swahili word for “truthful” is an NFT certification platform that relies on a novel Proof of Democracy consensus algorithm to enable its community to prove the validity of digital assets.
The way it works is fairly simple. Anyone, such as NFT creators, can request a certificate to validate an NFT by staking WAKU tokens. The protocol uses human certifiers, who also stake WAKU, to check the authenticity of an NFT and issue a certificate, earning yield on their stake for doing so. Then, the wider Wakweli community is free to challenge the validity of any certificate by staking WAKU tokens and providing evidence that shows a certified NFT is fake. The availability of various ML-based tools should ensure that both Certifiers and Challengers alike can find all the evidence they need to back up their claims. It’s a powerful protocol that incentivizes all players to act honestly, and as it becomes more widespread, and hopefully integrated with leading NFT marketplaces, incidences of fraud should become a lot less common.
The combination of machine learning and Wakweli’s strong incentives is desperately needed, because the NFT industry has long struggled to overcome the plague of scams. NFTs are a technology with almost limitless potential, but until everyday users have an easy way to verify the authenticity of digital assets, they’re not going to see widespread adoption.