AI Agents Can Already Pay in Crypto. Can a Website Tell Who They’re Working For?

AI agents can already make online payments using stablecoins, crypto tokens pegged to the U.S. dollar. However, a significant issue remains on the receiving end. When a website receives a request from an agent, it can’t determine which user the agent represents or whether that user approved the expenditure.

In response, some of the biggest names in U.S. commerce have opted to shut out these agents. Amazon (NASDAQ: AMZN) has blocked a new shopping agent developed by Meta Platforms (NASDAQ: META), while JPMorgan Chase (NYSE: JPM) has prohibited outside agents from accessing customer accounts. Meanwhile, Boston-based identity verification company Proof has introduced a protocol called x401 to reveal who an agent represents.

Crypto prices haven’t reacted to this move; XRP (CRYPTO: XRP) is trading at $1.31 as of October 15, 2026, down 4.6% in the last 24 hours, similar to Bitcoin’s (CRYPTO: BTC) 2.9% drop. Since agents can already manage transactions, what might motivate banks and major retailers to let them in?

Why Amazon and JPMorgan Are Blocking AI Agents

On September 8, Meta launched its personal AI agent, Muse, which can handle tasks such as booking travel, filling out forms, and shopping online. However, Amazon blocked Muse just two weeks later. Now, users encounter a warning indicating that access by an unauthorized AI agent violates Amazon’s terms of use.

Amazon claims Meta did not notify it of the integration, and Muse does not identify itself as an AI agent while browsing. Meta contends that Muse has no access to passwords or payment methods, and every transaction requires user approval. Nonetheless, these safeguards operate solely within Muse, leaving Amazon unable to verify who is actually making a purchase or if that individual has authorized the transaction.

Amazon previously won a preliminary injunction against another shopping assistant developed by Perplexity. However, the Ninth Circuit vacated that injunction on August 4, ruling that Amazon was unlikely to prove any breach of anti-hacking laws because the user accessed the site with the agent’s help. Despite this, the ruling still allows Amazon to regulate access under its terms of use, which is how Amazon is currently blocking Muse.

During JPMorgan’s earnings call on October 13, CEO Jamie Dimon stated the bank would not permit external agents to interact with customer accounts. “You don’t want rogue agents, and there could be thousands of them,” he expressed. He also highlighted concerns about a potential “liability transfer,” meaning customers could bear the costs of a faulty payment an agent makes. This restriction is more significant than Amazon’s because it prevents agents from accessing customer accounts for payments.

Shopify Is Letting AI Agents In as Bot Traffic Passes Humans

In contrast, some companies are embracing AI agents. Shopify (NASDAQ: SHOP | SHOP Price Prediction) has integrated Meta as an AI commerce channel and is expanding the partnership so that Muse can process purchases through Shopify’s payment system.

The movement toward AI agents is gaining momentum, with Cloudflare reporting that bots accounted for 57% of web-page requests on its network in June, surpassing human traffic for the first time. HUMAN Security noted an astonishing 7,851% increase in AI agent traffic in 2025. However, consumer opinions remain mixed: a survey found that 51% of 1,000 U.S. consumers are uneasy with AI tools storing their card details.

How x402 Moves AI Agent Payments, and x401 Verifies Identity

The payment process already exists, built on HTTP, the protocol that facilitates communication between browsers and websites. Code 402, long reserved for “payment required,” is now being used through a standard called x402, allowing agents to pay for web resources with stablecoins.

Coinbase (NASDAQ: COIN) initiated the development of x402, which the Linux Foundation now oversees through the x402 Foundation, whose members include Visa, Mastercard, and Stripe. However, according to Circle, x402 addresses how payments are made but not who the agent is.

Proof, formerly Notarize, says it is the first company to implement verified identity and transaction approval for AI agents. Their x401 protocol builds on HTTP code 401—the standard response for “unauthorized”—enabling a website to query an agent about whom they represent and what has been approved.

To verify the person behind the agent, Proof matches their identity against a government-issued ID and then requires them to authorize the transaction by signing it with a fingerprint or facial scan. For example, a person could confirm an agent’s $25,000 wire transfer through this method, and any modifications the agent makes afterward would invalidate the signature.

Proof charges websites for each verification over the x402 system, and Superfluid, a crypto settlement firm, was the first to use the service in September. “Blocking agents is understandable, but it’s not sustainable,” said Pat Kinsel, Proof’s CEO. Proof’s announcement doesn’t specify any particular blockchain or stablecoin, suggesting a neutral stance toward any single token.

Do AI Agent Payments Give XRP Holders Anything?

For now, agent identity gives XRP holders no direct reason to expect a higher price. Ripple says AI agents have made 13 million payments on the XRP Ledger using x402, about 25% of network activity. Those payments settle in stablecoins, though, so they measure ledger activity rather than demand for XRP.

If a major U.S. bank or retailer names an identity standard it would accept at Money20/20, which runs October 18 to 21, JPMorgan’s block could start to relax. Without one, the blocks will likely remain, and those 13 million payments will stay limited to sites that never check who the agent works for.

Contact [email protected] for any questions or corrections.