Artificial intelligence, the big shiny new thing that you either love or hate, is pretty much everywhere today; in search engines, organizing social media feeds, writing code and even dissecting medical scans. AI has no doubt become the obsession of every founder and investor, and it’s not just limited to Silicon Valley; every sector, whether mining, telecommunications, government or banking, is trying to understand what AI means for them. 

Which brings us to the question: What does AI mean for blockchains, and how can we find the intersection between these two worlds?

Bringing AI to Blockchains

It may come as a surprise to a few people, but AI isn’t just a clever chatbot that some people end up engaging with in oddly sci-fi ways. The technology itself is fantastic for creating autonomous actors that can reason, plan and make decisions. 

The Ethereum Foundation saw the potential of AI and recently revealed the launch of a dedicated decentralized AI (dAI) initiative. This initiative, led by EF’s Davide Crapis, wants to make Ethereum the preferred settlement and coordination layer for artificial intelligence.

At the heart of this effort is ERC-8004, a new proposed standard that enables participants to choose and discover AI agents without requiring any pre-existing trust. ERC-8004 provides a framework for AI agents to register onchain, attach verifiable metadata about what they are and how they operate, and build their reputation over time. 

The purpose of this standard is to guarantee that AI agents are who they claim to be and keep them accountable within the rules of a decentralized, open system. This is particularly important because these agents will be responsible for managing consequential assets, contracts and transactions on behalf of their participants. 

In the future, digital agents may play a larger role in our economies, and blockchains will become the neutral arbiter of their identity and trustworthiness.

Challenges of Building with AI

The road to bringing AI onto the blockchain won’t be entirely linear. For example, if thousands of agents began to transact in large numbers on Arbitrum, costs could rise, and blockspace may become clogged. Security is another issue to keep in mind, as AI can only launch in the blockchain space if malicious or misaligned agents are removed. 

These frictions are expected when building any new paradigm, and they must be tested, stressed and iterated on. Heavy AI workloads will almost certainly run offchain, which is why verifiable cryptographic proofs play an extremely important role if we wish to bring AI onchain. 

Arbitrum is committed to testing, studying and addressing some of these potential concerns in real-time, making sure that the chain is always prepared for what is next.

What does this mean for Arbitrum?

The first signs of dAI are already visible in the Arbitrum ecosystem. The recent Agent Arena showdown saw five different AI trading agents (each with varying trading strategies) compete for 30 days using onchain trading. Their actions were completely transparent, and outcomes were visible to anyone. 

In addition to Agent Area, projects funded by the Trailblazer AI Grant Program, including Allora, Eternal AI, NRN Agents, Ora and others, are all building the necessary frameworks for AI-driven prediction markets, signal inference and trading automation on Arbitrum. Allora, for example, is already working with PancakeSwap to power prediction markets where users can trade against or alongside machine-learning forecasts and Giza has recently launched its Pulse Agent on Arbitrum, adding another live example of AI-driven predictive analytics into the ecosystem.

Grantees alone are not the only ones making noise about dAI in and around Arbitrum. CapX, for example, is developing an “AI Agent Economy” chain. Agents will be able to build, own and exchange digital assets in an onchain marketplace. USD.AI is utilizing Arbitrum technology and AI to build a yield-bearing, synthetic dollar protocol. ZyFAI deploys automated DeFi agents designed to optimize yield on Arbitrum. GAIB is working on tokenizing GPU compute capacity, turning a scarce blockchain resource into a financial primitive using AI, and Kudai is building real-time capital optimization agents. 

All these different projects show that dAI in the Arbitrum ecosystem is not a single vertical or a side experiment; it takes on various layers from trading and prediction to infrastructure and building better user experiences. 

The Beginning of a Next Chapter

Beyond the surface-level hype around AI, something durable is forming. The intersection of blockchain and AI will be here to stay: from trading arenas, to prediction markets and tokenized compute, innovative projects on Arbitrum are already pushing the limits of what blockchains are capable of.

Interested in all the projects building AI in the Arbitrum ecosystem? Check out our Portal to discover the power of Arbitrum Everywhere.