Agentic AI might soon get into cryptocurrency trading — what could possibly go wrong?

Proponents of Web3 — what’s been called the next-gen internet — believe it could go mainstream this year by merging AI agents with blockchain to create secure, decentralized, and user-friendly digital experiences.

Agentic AI holds out the promise of simplifying complex tasks such as crypto trading or managing digital assets by automating decisions, enhancing accessibility, and masking technical complexity.

Web3 (or Web 3.0) envisions an internet where control and data shift from centralized companies to users through decentralized apps (dApps), cryptocurrency, smart contracts, and direct ownership of digital assets, such as non-fungible tokens (NFTs). AI agents promise to transform Web3 by increasing efficiency in trading, and by working alongside blockchain’s trusted security create advanced, self-governing applications.

It’s a combination that could revolutionize industries from finance to governance, art, and gaming, while also accelerating a shift to a user-empowered internet and Web3 apps. And by providing intuitive user interfaces and intelligent decision-making tools, those applications could become more widely accessible to users who won’t even need to understandithe underlying blockchain technology.

The two could, for example, simplify DeFi (decentralized finance) and smart contracts by automating functions — effectively opening the door to new users who may not have the technical chops to handle those tasks now. AI agents on blockchain can act independently to buy, sell, negotiate, or provide services on behalf of users.

Imagine a online wallet with an embedded AI agent that scouts out the best DeFi yields, makes swaps, or participates in Decentralized Autonomous Organization (DAO) — no human input needed. Or imagine insurance payouts triggered not just by oracles (rules-based embedded code) but by AI risk analysis in real time.

Developers could share or rent their AI tools, and users can access decentralized AI without relying on Big Tech companies. Think of it like an “App Store” for AI but Web3-native, peer-to-peer, and without intermediaries. Blockchain also brings auditability to black-box models, especially important in fields like healthcare, finance, or law.A user could verify if an AI made a loan decision fairly, because all steps are traceable on a public ledger.

Tools for Humanity and its World project

One effort under way for the past two years involves the non-profit Tools for Humanity’s World project, which has features like the World App and biometric identity verification that purports to offer users secure, human-verified access to financial services and decentralized dApps. While some experts see the integration as a potential breakthrough, others remain skeptical, citing privacy concerns, immature AI autonomy, and limited real-world utility as barriers to mass adoption.

At the moment, cryptocurrency power users have skillsets that allow them to easily swap tokens and understand slippages, liquidity, and pool dynamics — they know how their trades affect the market. But an AI agent could open the door to non-experts, giving them access to these kinds of financial opportunities by simplifying the process, allowing them to make time-sensitive trades when the time is right.

“So, [agentic AI] being able to act on your behalf, can make a trade on your behalf, for example. It’s another kind of a use case for reducing complexity, as well as empowering Web3 apps to do more,” said Steven Smith, head of protocol and applied research at Tools for Humanity, the nonprofit co-founded by OpenAI CEO Sam Altman. Tools for Humanity develops Web3 technology and is working to empower individuals with secure, private, and decentralized digital identity and transaction solutions.

But if AI agents are to be added to the mix of blockchain-based applications, solutions will be needed to ensure fairness and human control.

In 2022, the organization launched its World project (previously known as Worldcoin), creating a global identity and financial network to enhance online trust and provide equitable access to the digital economy.

Key components of the project include:

World App: A digital wallet for identity authentication and financial transactions.

World ID: A digital passport to verify humanness and uniqueness online, distinguishing humans from AI tools.

Orb: An Iris-scanning device for secure, private identity verification. Once the iris image is on the blockchain ledger, it can verify a human being from a bot.

In particular, blockchain combined with AI tools could help ensure Web3 transactions are more secure from “Syble attacks” — the ability of one person to impersonate many others, in some cases thousands of people, through duplicative online accounts.

The World App is a mobile application developed by World, a global project that aims to create a new economic system based on a decentralized, global currency.
Tools for Humanity

The World Project already uses blockchain in the World Chain network it relies on to link its app and other tools in several ways, including:

Decentralized Finance (DeFi): Enabling access to financial services without traditional banks.

Digital Identity: Creating secure, verifiable identities to access services like education and healthcare.

Social Impact Projects: Supporting transparent funding and tracking for global challenges.

Community Building: Connecting people globally to collaborate and share resources.

Over the past two years, the project has expanded globally. This year, it announced it has over 10 million users across 160 countries conducting more than 115 million transactions. It has also faced greater regulatory scrutiny concerning biometric data processing, leading to commitments to allow users to delete their data permanently in light of privacy concerns.

“We’re the second largest consumer of theory of L1 resources from a data storage perspective. So it’s an incredibly active chain,” Smith said.

Agentic AI and proof of humanity

However, like many online systems, “it could also be very, very abused. But by integrating a proof of humanity function in there, you can ensure that you know that agent is operating on behalf of a specific person who has empowered it.”

Using AI agents is key, Smith said. In October, Tools for Humanity conducted a successful beta test of its AI-enabled Web3 network with blockchain platform provider Polygon. It then migrated its World ID and World App to Polygon’s mainnet and launched a “Mini App” store within the World project.

“That allows users to come in and easily develop applications that run in this environment that can take advantage of a proof of human [technology],” Smith said. “We’ve got multiple mini apps…and we’ve developed a very simple interface that allow a user to make trades if they want to trade one token for another token.”

The World App offers Mini Apps as savings vehicles where users can earn interest on token deposits, games that reward players with in-game tokens, and a popular app called World Chat, an encrypted messaging service similar to Signal — all within World App, Smith said.

The World project also recently partnered with Razor to help fight third-party bots in online games amid the rise of AI gaming by verifying user IDs.

Adrian Leow, a Gartner vice president analyst and lead author of the “Hype Cycle for Blockchain and Web3,” said the intersection of advanced technologies like agentic AI, quantum computing services and blockchain will be the required catalyst to drive blockchain forward into mainstream scalable adoption.

World Orb is a device used by World, the Web3 project. It’s a key component in their biometric identity verification process. The World Orb is designed to scan a person’s iris (eye) to uniquely identify individuals without collecting other personal information. This biometric scanning helps Worldcoin ensure that each participant is unique, preventing multiple sign-ups by the same person.
Tools for Humanity

Blockchain, Leow said, has gone through many ups and downs – “crypto winters, resurgences like NFTs, and then declines.

“It’s been a roller coaster…that hasn’t yielded long term scalable benefits for enterprises on its own,” he said. “When you combine both [blockchain and agentic AI], you could get automated agents operating on decentralized infrastructure. This combo potentially unlocks several efficiency and trust benefits.”

While use cases, such as cryptocurrency, DeFi, NFTs, and dApps show blockchain’s potential, the core technology itself still holds “revolutionary promise.” It’s important to focus on the technology, not just “the hype,” Leow said.

In Web3, Leow siad, agentic AI could be used to:

Automate dApp interactions: Instead of manually interacting with DeFi, NFTs, or DAOs, AI agents could handle transactions, monitor opportunities, rebalance portfolios, or vote on proposals.

Serve as autonomous economic actors: They could earn, spend, or stake tokens based on pre-programmed goals or learned strategies.

Act as user proxies: Represent users across the decentralized web, negotiating, buying, and selling digital goods (NFTs, services, etc.) on their behalf.

Blockchain, in turn, brings transparency, immutability, and decentralization, which makes it ideal for trustless environments, according to Leow. In Web3 it can be used to track transactions publicly and securely; allows for smart contracts that let parties exchange value without intermediaries; and it enables ownership and provenance of digital assets.

A powerful tech combo that still needs to mature

Not everyone is bullish on the intersection of Web3, agentic AI and blockchain. Forrester Research vice president and principal analyst Martha Bennett is among those who are skeptical. In 2023, she co-authored an online post critical of Worldcoin, now the World project, and her opinion hasn’t changed in several regards.

World project still faces major challenges, including privacy issues and concerns about its iris biometric technology, she said. And Agentic AI is still in its early stages and not yet capable of supporting Web3 transactions. Most current generative AI (genAI) tools, including LLMs, lack the autonomy defined as “agentic AI.”

“There’s no AI technology today that would be able automate Web3 transactions in a reliable and secure manner,” she said.

Given the risks and the potential for exploitation, it’s too soon to rely on AI systems with high autonomy for Web3 transactions. She did note, however, that Web3 already uses automation through smart contracts — self-executing electronic contracts with the terms of the agreement directly written into code.

“Will Web3 go mainstream in 2025? My overall answer is no, but there are nuances,” she said. “If mainstream means mass consumer adoption, it’s a definite no. There’s simply not enough utility there for consumers.”

Web3, Bennett said, is largely a self-contained financial ecosystem, and efforts to boost adoption through Decentralized Physical Infrastructure Networks (DePIN), such as Tools for Humanity’s, haven’t led to major breakthroughs. Even so there has been a shift in how the financial sector views public blockchain networks, with increased exploration under way, she said.

Jack Gold, principal analyst with tech industry research firm J. Gold Associates, said the first issue around transactions, “AI or not,” is that they must be completely secure.

“While many believe that blockchain is a secure method of doing transactions, it has been broken for some time – it’s how many malware/ransomware transactions were retrieved by government agencies after companies paid the ransom,” he said. “So while AI agents may make the mechanics of transactions easier, blockchain may only be sufficient for non-really critical transactions.”

And, Gold said, AI can also be turned into a “bad actor” by playing with the training parameters to include things that should not be able to happen. (If you train it properly, an AI agent might think it perfectly normal to inform someone else of a user’s secure transactions, as an example.)

Smith argued that the World project’s ability to prove a Web3 user is human thwarts deepfakes, scams, misinformation and agent swarms.

“We just ensure that [Web3 is] able to be used in an efficient and safe manner. So, when you look at like AI, it is very powerful in general, everyone knows that,” Smith said. “Then, when you look at agentic AI, you’re able to operate in a Web3 environment. It could potentially be very powerful.”