Amazon v. Perplexity: The Court Case That Will Define Who Owns the AI Shopping Layer
On April 1, 2026, Perplexity AI filed a 96-page opening brief with the United States Court of Appeals for the Ninth Circuit, asking the court to vacate an injunction that bars its Comet AI shopping agent from accessing Amazon. Case number 26-1444 will determine whether a 1984 federal hacking statute can be weaponized to block AI agents that consumers voluntarily install on their own devices.
This is not a corporate spat between two tech companies. This is the legal fight that will define the boundary between platform control and consumer autonomy in the age of agentic commerce. Every retailer, every AI company, and every brand building for the post-search economy has a stake in the outcome.
The Core Dispute: Who Authorizes an AI Agent?
The case centers on a deceptively simple question: when a user logs into their own Amazon account and tells an AI assistant to shop for them, is that access "authorized" under federal law?
Amazon says no. The platform argues that its Conditions of Use require AI agents to identify themselves (via user-agent strings) and restrict agent access to public portions of the site. Perplexity's Comet browser violated those terms by accessing logged-in user accounts without identifying itself as an AI agent. Amazon sent a cease-and-desist letter, implemented technical measures to detect and block Comet's digital fingerprint, and ultimately filed a complaint in November 2025 under the Computer Fraud and Abuse Act.
Perplexity says yes. Users voluntarily installed Comet, logged into their own Amazon accounts with their own credentials, and explicitly directed the AI assistant to browse and purchase on their behalf. Perplexity's servers never connect directly to Amazon's computers. The data flows from Amazon to the user's browser first, then encrypted screenshots are sent to Perplexity for processing and deleted within 30 days. Amazon's own expert conceded there are "no direct requests from Perplexity hosts to Amazon.com."
U.S. District Judge Maxine M. Chesney sided with Amazon on March 9, 2026, granting a preliminary injunction. The Ninth Circuit granted a temporary stay on March 16, keeping Comet operational while the appeal proceeds.
Why This Case Matters: $69 Billion in Ad Revenue at Stake
Strip away the legal jargon and the core issue becomes financial. Amazon's advertising business generated $68.6 billion in 2025, with Q4 alone reaching $21.3 billion at 23% year-over-year growth. Approximately 90% of that revenue comes from Sponsored Products and Sponsored Brands: the ads users see when they search for items on Amazon.
An AI shopping agent does not browse. It does not scroll past sponsored products. It does not get distracted by recommendation algorithms. It finds what the user asked for and facilitates the purchase. In that transaction, advertising ceases to have a function.
Amazon's internal concern is explicit. The appeal brief alleges that Amazon's true motivation is to identify and block AI agents so it can force "customers to view its advertisements." Amazon's own documents reveal that advertisers do not pay for their ads to be shown to automated agents.
This is the same structural tension that Google faces with AI Overviews cannibalizing search ads, that publishers face with AI engines consuming their content without sending traffic, and that the entire digital advertising industry faces as AI intermediaries stand between consumers and content. Amazon is simply the first major platform to fight it in court.
The Legal Architecture: CFAA, Van Buren, and hiQ v. LinkedIn
The legal arguments in this case build on two landmark decisions that fundamentally shape how the Computer Fraud and Abuse Act applies to web access.
Van Buren v. United States (2021) established the Supreme Court's "gates-up/gates-down" test. The Court held that accessing areas of a system you are already authorized to access, even for an improper purpose, does not constitute a CFAA violation. A violation requires bypassing an explicit barrier such as a password or technical block. The Court specifically warned against reading the CFAA to criminalize "every violation of a computer-use policy," noting this would extend "criminal penalties to a breathtaking amount of commonplace computer activity."
hiQ Labs v. LinkedIn tested whether platforms can use the CFAA to block scraping of publicly available data. The Ninth Circuit held that a "defining feature of public websites is that their publicly available sections lack limitations on access." LinkedIn's cease-and-desist letter and terms of service were not sufficient to create CFAA liability for accessing public data.
Amazon's case introduces a critical variable: password-protected accounts. Unlike hiQ, where the data was publicly accessible, Amazon's user accounts sit behind authentication gates. The district court found that the Facebook v. Power Ventures precedent applied: once a platform explicitly revokes authorization, any future access becomes unauthorized under the CFAA.
But the IAPP's legal analysis identifies a nuance the district court may have overlooked. Facebook's cease-and-desist letter "explicitly and unequivocally revoked Power Ventures' authority to access its site." Amazon's cease-and-desist letter did not revoke Perplexity's access entirely. It specifically left open the door for Perplexity to continue accessing the platform, provided it did so "transparently" and in accordance with Amazon's Conditions of Use. As the IAPP analysis concludes: "At best, Perplexity is 'exceeding authorized access' by failing to identify itself transparently, but that is conditioned on defying Amazon's Conditions of Use, which returns us squarely to a contractual dispute" that Van Buren held is not enough to invoke the CFAA.
Amazon's Hypocrisy Problem: Rufus and "Buy for Me"
Perplexity's appeal brief raises a devastating counterpoint that has received insufficient attention in media coverage.
While suing Perplexity for deploying an AI shopping agent on Amazon's platform, Amazon simultaneously operates its own AI shopping agents on other platforms. Amazon's Rufus AI assistant reached more than 300 million users throughout 2025 and generated nearly $12 billion in incremental annualized sales. Rufus can track prices, monitor deals, and complete purchases autonomously through an auto-buy feature.
More damaging: Amazon's "Buy for Me" feature enables Amazon's AI to purchase items from third-party retailer websites, often without those retailers' knowledge or permission. Products available through Buy for Me grew from 65,000 at launch to over half a million. The appeal brief claims Amazon even listed Perplexity's own merchandise through Buy for Me without Perplexity's consent.
The brief frames this as an unclean hands argument. Amazon seeks to block Perplexity from doing on Amazon exactly what Amazon does to every other retailer through Buy for Me. If the Ninth Circuit finds this persuasive, it undermines Amazon's equitable position significantly.
The $15 Trillion Domino Effect
The implications of this ruling extend far beyond Amazon and Perplexity. Gartner projects that by 2028, 90% of all B2B purchases will be handled by AI agents, channeling more than $15 trillion in spending through automated exchanges. That projection assumes AI agents can access platforms on behalf of users.
If Amazon wins on appeal, the precedent establishes that platforms can use terms of service to block third-party AI agents entirely. The domino effect is immediate:
- Retailers follow Amazon's playbook. The Wall Street Journal reports that retailers across the industry are already reading the district court ruling as a signal they can wall off AI agents. Expect updated terms of service blocking AI agent access across major e-commerce platforms within months.
- Agent interoperability dies. Protocols like OpenAI's Agentic Commerce Protocol and Google's Universal Commerce Protocol assume AI agents can access merchant systems. If platforms can unilaterally block access, these protocols become meaningless.
- Platform-owned agents win by default. If only Amazon's Rufus can shop Amazon, only Google's Gemini can navigate Google Shopping, and only each platform's native AI can access its inventory, the agentic commerce future becomes a collection of walled gardens. Consumers lose the ability to use a single AI agent across platforms.
- Structured agent access becomes the standard. Rather than blocking AI agents, platforms would need to build proper access protocols, permissions systems, and agent-readable data formats. This accelerates the adoption of protocols like ACP and UCP.
- Advertising models adapt. If AI agents cannot be blocked, Amazon and other platforms must develop new advertising formats that work within AI-mediated discovery. This could mean sponsored recommendations within AI agent results, product placement in agent responses, or transaction-based advertising models.
- Consumer agency expands. Users retain the right to delegate shopping to AI agents of their choice, creating genuine competition in the AI commerce assistant space.
The Amicus Brief Signal
The number and diversity of amicus curiae briefs filed in this case signal its significance. Multiple industry groups, legal organizations, and technology companies have weighed in. This is unusual for a preliminary injunction appeal, and it indicates that the legal community recognizes this case as potentially precedent-setting.
The briefs address fundamental questions that extend beyond Amazon and Perplexity:
- When a user delegates a task to an AI agent, does the agent inherit the user's access rights?
- Can terms of service override a user's affirmative decision to authorize a third-party agent?
- Should the CFAA, designed to punish computer hacking in 1984, govern AI agent behavior in 2026?
- Who bears liability when an AI agent violates a platform's terms: the user who directed it, the company that built it, or both?
What the Ruling Means for Brands and Marketers
Regardless of how the Ninth Circuit rules, the implications for brands building AI visibility strategies are clear.
If platforms can block AI agents: Your product data must be optimized for each platform's native AI assistant. Rufus optimization becomes as important as Amazon SEO. You cannot rely on third-party AI agents to discover and recommend your products across platforms. Each platform becomes its own AI discovery silo.
If AI agents roam freely: Your product data must be structured, comprehensive, and machine-readable. AI agents will evaluate products across platforms simultaneously, comparing prices, reviews, and specifications. Structured data (schema markup, clean product feeds, accurate availability data) becomes the single most important factor in AI agent discovery. Brands with incomplete or outdated product information become invisible.
In both scenarios, the winning strategy is the same: make your product data impeccable.
Ensure product titles, descriptions, and metadata are accurate and comprehensive. Implement structured data across all platforms where your products appear. Maintain real-time inventory and pricing data. Build the kind of authoritative brand presence that AI agents trust for recommendations, whether those agents are platform-native or third-party.
Timeline and What Happens Next
The appeal process has several key milestones:
- April 1, 2026: Perplexity filed its 96-page opening appeal brief
- Administrative stay in effect: Comet remains operational during the appeal
- Amazon's response brief: Expected within 30-60 days
- Oral arguments: Likely late Q2 or Q3 2026
- Ninth Circuit decision: Probably Q3-Q4 2026
The timeline matters because the agentic commerce market is not waiting for courts. Shopify activated Agentic Storefronts for all eligible merchants on March 24. Walmart launched advanced ChatGPT shopping discovery on March 30. Carrefour became the first major European retailer to integrate full product catalogs into ChatGPT. The technology is deploying faster than the law can adapt.
The Bigger Picture: Platform Power in the AI Era
Zoom out further and Amazon v. Perplexity is the first battle in a much larger war over who controls the discovery layer of commerce.
For two decades, Google controlled discovery through search rankings. Brands spent billions on SEO to be visible in Google's results. Now AI agents are becoming the discovery layer, and platforms are fighting to ensure that only their own AI agents can access their inventory.
Amazon's position is rational. If third-party AI agents bypass advertising, Amazon's most profitable business line erodes. But the consumer argument is equally rational: people should be able to choose which AI assistant shops for them, just as they can choose which browser to use.
The Ninth Circuit does not need to resolve the philosophical question. It needs to decide whether the CFAA, written before the internet existed, applies to AI agents acting at a user's direction on password-protected platforms. The answer will shape the architecture of commerce for the next decade.
How Visible Is Your Brand to AI?
88% of brands are invisible to ChatGPT, Perplexity, and Gemini. Find out where you stand in 60 seconds.
Check Your AI Visibility Score FreeFrequently Asked Questions
What is the Amazon v. Perplexity case about?
Amazon sued Perplexity AI under the Computer Fraud and Abuse Act (CFAA) for allowing its Comet AI shopping agent to access users' password-protected Amazon accounts without identifying itself as an AI agent. Perplexity argues that users voluntarily authorized Comet to shop on their behalf, making the access lawful. The Ninth Circuit Court of Appeals will decide whether platform terms of service can override user consent to AI agent access.How does this case affect the future of agentic commerce?
The ruling will set the first major legal precedent for whether platforms can block third-party AI agents. If Amazon wins, every major platform could adopt similar blocking strategies, creating walled gardens where only native AI assistants can operate. If Perplexity wins, structured agent access protocols become the industry standard and consumers retain the right to choose their AI shopping assistant.Why is Amazon's advertising business relevant to this lawsuit?
Amazon generated $68.6 billion in advertising revenue in 2025, primarily from Sponsored Products that users see when browsing. AI shopping agents bypass these ads entirely: they search, compare, and purchase without scrolling past sponsored listings. If AI agents cannot be blocked, Amazon must develop entirely new advertising models for AI-mediated commerce.What is the "unclean hands" argument in Perplexity's appeal?
Perplexity's appeal brief argues that Amazon operates its own AI shopping agents on other platforms through features like "Buy for Me," which purchases products from third-party retailers without their knowledge or permission. Amazon cannot seek equitable relief while engaging in equivalent conduct against competitors, according to Perplexity's legal team.When will the Ninth Circuit issue its ruling?
The appeal process typically takes several months. Amazon's response brief is expected within 30-60 days of Perplexity's April 1 filing. Oral arguments are likely in late Q2 or Q3 2026, with a decision probable in Q3-Q4 2026. The administrative stay means Comet remains operational during the appeal.How Visible Is Your Brand to AI?
88% of brands are invisible to ChatGPT, Perplexity, and Gemini. Find out where you stand in 60 seconds.
Check Your AI Visibility Score Free