In the future, robots will help you buy, buy, buy, but disputes will become a problem: A cryptocurrency project launches an Internet court to quickly resolve agent business conflicts.

Written by: Forbes

Translated by: AididiaoJP, Foresight News

In the not-so-distant future, everyone will use artificial intelligence agents. These agents connect to the cloud, reside in mobile phones, and can help you reply to emails, book flights, and even automatically perform tax-loss harvesting for investment portfolios.

Robinhood clients are now using AI agents to analyze stock market fluctuations and autonomously trade based on personal instructions. SAP's Joule helps business clients analyze inventory, find the best suppliers, and complete purchases. Shopping agents like Amazon's "Buy for Me" scan online for the best deals at machine speed, negotiate terms with sellers' agents, set delivery times, and make direct payments.

Well-known AI and cryptocurrency companies, from Anthropic and OpenAI to Coinbase and Circle, are pushing for this agent-driven future to arrive as soon as possible.

But what happens when the color of the sofa ordered by the agent is wrong, it arrives two weeks late, or the goods are damaged and the seller insists it was an issue after delivery? This seemingly insignificant but potentially costly issue lurks within the grand blueprint of agent-based commerce. Although software can now represent humans and businesses in shopping, bargaining, hiring, and paying, AI sometimes experiences "hallucinations," and commerce is never just a monetary exchange; accidents will always happen.

"Agent-based commerce has reached a critical turning point, yet we are not prepared for the potential consequences," said David Riudor, CEO and co-founder of GenLayer Foundation. The foundation, based in the Cayman Islands, operates a new blockchain called GenLayer and its core application—the Internet court. This court is specifically designed to adjudicate disputes between agents, operating without human intervention and has gained support from 26 cryptocurrency and AI companies, including the cryptocurrency exchange OKX, wallet provider MetaMask, and Binance's BNB chain.

The Internet court does not aim to completely replace human judges with robots but rather helps agents pre-sign contracts with clear terms. When the two parties cannot reach an agreement on the outcome, an AI jury evaluates the evidence and provides a ruling within minutes.

Riudor pointed out that this technology is particularly suitable for small transactions—hiring a lawyer is not economical, but ignoring the issue could lead to a loss of costs. "We are not competing with the traditional legal system," said Albert Castellana, co-founder and CEO of GenLayer Labs, "we are simply providing an alternative. For a $10,000 claim, hiring a lawyer is not cost-effective; this system can resolve it quickly, potentially costing just a few cents."

The potential market size is enormous. According to data from Adobe Analytics, since October 2024, the traffic to retail websites driven by AI recommendations has increased over 14 times. McKinsey predicts that by 2030, AI agents could facilitate between $30 trillion to $50 trillion in consumer business transactions worldwide. However, most emerging infrastructure is still focused on the "smooth path"—agents find what the owner wants, complete payment, receive goods or services, and move on.

Currently, the Internet court is applied in limited scenarios. The social platform Collective Memory rewards users for uploading real-time photos, videos, and reports, and when there is a need to determine whether a controversial image is fake, it calls on the GenLayer system. For example, for video from a bombed school in Gaza or scenes from the streets of Tehran, the Internet court assesses authenticity based on uploaded time, location, relevant submission records, and user history activities.

The ultimate goal is to allow the Internet court to automatically intervene when AI agents dispute with each other.

Imagine a small online clothing company where the owner has delegated daily operations to multiple agents: one manages inventory, another buys advertising, and another finds creative solutions. The owner wants a new logo, and the agents find another designer represented by an agent. The two parties agree on the design, price, and delivery date; the logo looks good upon delivery, but is later found through a reverse image search to possibly have plagiarized someone else's portfolio.

The design of the Internet court is precisely aimed at situations like this: allowing agents to pre-agree on terms, escrow payments, and submitting disputes to the jury for processing before funds are released.

The core of the jury is blockchain technology. It consists of 5 randomly selected blockchain validators, each running different models (such as Claude, GPT, Gemini). One of them acts as a leader and proposes an initial decision, while the others initially vote secretly, then publicly state their agreement. If consensus is reached, a 30-minute dispute window opens, where agents or humans can challenge by paying a deposit. If challenged, the jury expands to 11 members, and this continues until consensus forms and no one raises objections.

This mechanism draws on the jury theorem proposed by the Enlightenment philosopher and mathematician Nicolas de Condorcet in 1785: the more independent evaluators there are, the higher the probability of arriving at the correct conclusion. GenLayer believes that a combination of multiple models is harder to manipulate than a single model or a single human arbitrator.

While disputes between agents might sound futuristic, the Internet court is already online and in testing. According to Castellana, the network processes about 350,000 transactions per day, generating 20,000 to 25,000 decisions. It aims for a formal public launch later this year, along with the issuance of tokens to attract more validators—anyone can take part in this role.

Riudor stated that the system is not limited to agent-based commerce in the future but can also be applied to prediction markets. For example, Polymarket currently relies on the UMA protocol to submit dispute outcomes to token holders for voting, whereas AI-assisted arbitration would be faster. "We are in discussions with some large prediction markets," Castellana said, "they are waiting for our full launch but are already evaluating."

Andrew Hall, a professor at Stanford University's business school and research advisor for the a16z crypto team, pointed out earlier this year that using large language models as adjudicating judges can help prediction markets scale because models cannot be bribed and performance is rapidly improving. But he also cautioned that models can still produce hallucinations and may be manipulated through clever prompts or contaminated training data.

Lindsay Lin, former legal advisor and current COO of the New York cryptocurrency venture capital firm Dragonfly, also sees similar issues: "Many large language models share training data and common flaws, which leads to correlations, while human judgment is often more independent. However, people tend to use AI for low-value disputes because it is cheaper and faster than human jurors. Agent-based commerce is vast, and the volume of disputes will also be large. It makes sense for agents to have standardized agreements so they clearly understand cooperation terms and remedies when transactions are improperly completed."

Others have reached similar conclusions. Just two weeks ago, the world's largest arbitration agency—the American Arbitration Association International Dispute Resolution Center—announced the launch of a "Legal Context Agreement" standard for agents. This agreement is co-led with Denver-based blockchain company Integra Ledger, with founding contributors including Google, IBM, and several leading cryptocurrency companies like Circle and Ava Labs.

Of course, whether these standards can be widely adopted ultimately depends on broad adoption and whether AI models can effectively reduce hallucinations and biases.

Meanwhile, the infrastructure for agents to search, hire, and pay for each other is rapidly taking shape. In recent weeks, GenLayer partner OKX and AI-focused NEAR blockchain team launched markets one after another, enabling agents to hire other agents for paid tasks, such as acquiring datasets or assisting with code reviews.

Real courts are also beginning to handle cases of AI agent violations. In November 2025, Amazon sued Perplexity, alleging that its AI-driven Comet browser logged into customer accounts, masquerading as a regular Google Chrome browser, making unauthorized purchases on the Amazon platform. In March, a federal judge in California issued a preliminary injunction preventing Comet from shopping on Amazon, but an appellate court later stayed that order, currently reviewing Perplexity's appeal.

Regardless of how the court ultimately rules, this case highlights a larger challenge facing agent-based commerce: When millions of AI agents act on behalf of users across platforms, how can effective regulation be ensured without a unified enforcement mechanism?