A federal judge has blocked Perplexity AI’s shopping bot from making purchases on Amazon, marking what may be the first significant legal confrontation over agentic AI commerce. The ruling, issued by a U.S. District Court judge, grants Amazon a temporary restraining order against Perplexity’s “Buy with Pro” feature — an AI agent designed to browse Amazon’s marketplace, select products, and complete transactions on behalf of users without ever opening Amazon’s app or website.
This isn’t a minor skirmish. It’s a direct collision between two competing visions of how consumers will shop online in the coming years.
As GeekWire reported, the case centers on Amazon’s argument that Perplexity’s agent violated its Terms of Service by automating interactions with the platform — scraping product data, navigating checkout flows, and placing orders through methods Amazon never authorized. Amazon contended that Perplexity was essentially building a business on top of Amazon’s infrastructure without permission, siphoning away the direct customer relationship that Amazon guards fiercely.
Perplexity launched “Buy with Pro” as a premium feature for its paying subscribers, pitching it as a frictionless way to shop. Tell the AI what you need, and it handles the rest: finding the product, comparing prices, and completing the purchase. The pitch was compelling. But Amazon saw it as parasitic.
The judge sided with Amazon, at least for now. The temporary restraining order prevents Perplexity from using its AI agent to interact with Amazon’s platform while the broader case proceeds. The court found that Amazon demonstrated a likelihood of success on its claims, particularly around violations of the Computer Fraud and Abuse Act (CFAA) and breach of contract through Terms of Service violations. The judge also noted potential irreparable harm to Amazon’s customer relationships and data control.
Why this matters for every company building AI agents
This ruling sends a signal far beyond the Amazon-Perplexity dispute. Dozens of startups and major tech companies are racing to build AI agents that act on behalf of users — booking flights, scheduling appointments, making purchases, filing forms. OpenAI’s Operator, Google’s Project Mariner, and a growing list of smaller players are all pushing toward a future where AI agents interact with websites and services autonomously. This case suggests that the platforms being acted upon aren’t going to sit quietly.
The core legal question is deceptively simple: Can an AI agent, acting on a user’s behalf, access a third-party platform in ways that violate that platform’s terms of service? And if the user themselves could perform the same actions manually, does delegating those actions to a bot change the legal calculus?
Amazon’s position is clear. Its Terms of Service prohibit automated access. Full stop. The company argues it doesn’t matter that a human initiated the request — the actual interaction was performed by software, and that’s what the terms prohibit. Perplexity countered that its agent was acting as the user’s delegate, essentially an extension of the user’s own intent and authorization. The judge wasn’t persuaded by that argument at this early stage.
Legal experts have noted the parallels to earlier web scraping cases, particularly hiQ Labs v. LinkedIn, which dealt with automated access to publicly available data. But the Amazon-Perplexity dispute involves something different: not just reading data but transacting. Placing orders. Spending money. That escalation makes courts more cautious, and gives platforms stronger arguments about fraud risk, liability, and consumer protection.
There’s also a business model question lurking underneath the legal arguments. Amazon’s entire retail operation depends on controlling the customer experience — from product discovery through checkout to post-purchase support. If AI agents can bypass that experience, Amazon loses its ability to upsell, cross-sell, display ads, and collect behavioral data. That’s not just a legal concern. It’s an existential commercial threat.
Perplexity CEO Aravind Srinivas has been vocal about his vision for AI-powered commerce, arguing that consumers should be able to delegate routine purchasing decisions to intelligent agents. In previous public statements, Srinivas has framed this as a natural evolution of search — from finding information to taking action. But the ruling suggests that “taking action” on someone else’s platform requires that platform’s cooperation, or at least its acquiescence.
So where does this leave the agentic commerce space? Uncertain, but not dead. The ruling is a temporary restraining order, not a final judgment. The full case will play out over months, possibly years. And the outcome could vary significantly depending on how the CFAA claims are interpreted and whether courts view Terms of Service as sufficiently binding to block agent-based interactions.
Some industry observers expect this to accelerate the development of formal agent-access protocols — essentially APIs or standards that allow AI agents to interact with platforms in authorized ways. If platforms like Amazon won’t allow unauthorized bots, the alternative is negotiated access. That could mean revenue-sharing agreements, certified agent programs, or new technical standards governing how AI agents authenticate and transact.
But don’t expect Amazon to open its doors willingly. The company has spent decades building a walled garden optimized for direct consumer engagement. Letting third-party AI agents mediate that relationship undermines the entire model.
For founders building in this space, the message is stark: if your AI agent touches a platform you don’t control, you need a legal strategy as much as a technical one. Terms of Service are enforceable. Courts will issue injunctions. And the biggest platforms have the legal resources to fight.
This case is early, but it’s already shaping the rules of engagement for the next era of AI-driven commerce. The technology may be ready. The law — and the incumbents — aren’t necessarily on board.
WebProNews is an iEntry Publication