Amazon vs Meta: Why ZK Cryptography Could Secure the Future of AI Commerce

ZK Cryptography: A Solution for Secure Agentic Commerce

Amazon’s blockade of Meta’s Muse AI agent over concerns that it was capturing and storing customer credentials spotlights zero-knowledge (ZK) cryptography as a promising solution for secure agentic commerce, industry experts say. The friction comes as autonomous shopping agents are poised to surge, with JPMorgan Chase forecasting that Muse AI could become a dominant force in the space.

Sachi Kamiya, director of venture and growth at Sentient Foundation, told Bitcoin.com News that ZK techniques do more than shield user credentials; they also offer protection against security risks such as prompt injection. As AI agents interact with an increasing number of active user accounts, mitigating prompt injection becomes a critical safety requirement.

Reports indicate that Amazon’s Sept. 20 decision to block Meta’s Muse AI followed unsuccessful attempts to persuade Meta to voluntarily exclude the platform. At the heart of the standoff was Muse AI’s practice of browsing and capturing user credentials without identifying itself, a risk Amazon warned poses significant privacy and security concerns. While the two tech giants are reportedly negotiating a resolution, experts see the conflict as a compelling case for ZK cryptography.

“In the context of AI agents, ZK allows an agent to verify a user’s identity, permissions or credentials without exposing those sensitive inputs to the AI model or a third party,” Kamiya explained. She added that the technology is critical for industries such as health care, finance and legal services, where agents must execute tasks using private data without unnecessarily broadcasting it.

Cham Ho, product lead at AIR and Moca Network, echoed these views, calling sitewide blocks on AI agents counterproductive.

“Blocking agents won’t stick long term. Even if platforms opt to block them, consumer demand for personalized, cross-platform assistants will drive developers to find workarounds. The winning strategy isn’t building walled gardens; it’s winning over agents with personalized, agentic-first offers.”

Experts Warn Against Blocking AI Agents

Beyond tech executives, regulatory figures are also weighing in. Outgoing Securities and Exchange Commission (SEC) Commissioner Hester Peirce recently endorsed the framework at SIFMA’s Digital Assets Conference in New York, arguing that traditional know-your-customer (KYC) and anti-money-laundering (AML) models rely on a flawed foundation, whereas ZK cryptography provides a superior, privacy-first path forward.

While Amazon frames the block as a move to safeguard customer privacy, critics view it as an effort by a big tech incumbent to protect its data moat and stifle competition. Others warn that the standoff marks the beginning of a broader trend toward closed, gated e-commerce ecosystems.

Beyond the immediate platform friction, some experts believe the dispute signals a deeper question about the future structure of the agentic economy. When asked whether rising agent adoption will trigger a wave of closed, gated ecosystems or whether open standards will prevail, Kamiya said the industry is at a pivotal moment.

She noted that early internet networks in the 1960s lacked universal standards until researchers developed shared protocols that allowed disparate systems to communicate — a shift that ultimately enabled the modern internet. AI agents, she said, are now entering a similar phase.

“We’re already seeing early examples, such as x402, which creates a standard for how agents can make payments,” Kamiya said. “Many more standards will emerge around identity, permissions, communication, security and data access.”

Kamiya argued that the agentic economy will thrive on network effects, with agents becoming far more valuable when they can operate across social platforms, marketplaces, databases, financial services and other applications rather than being confined to a single closed ecosystem.

“For that open ecosystem to exist, we need common standards and rules that different platforms and agents can follow,” she said. “Open standards will ultimately prevail because connected ecosystems create and accrue more value than isolated ones.”

She added that this vision aligns with Sentient Foundation’s mission to build open-source artificial general intelligence (AGI) and the frameworks needed for an interoperable agent ecosystem, rather than a future dominated by closed, proprietary platforms.