In a significant escalation of tensions between e-commerce incumbents and artificial intelligence innovators, Amazon has formally moved to block the AI-powered shopping assistant from the startup Perplexity. The online retail giant issued a cease-and-desist letter and subsequently filed a lawsuit, demanding that Perplexity’s browser extension, Comet, be barred from making automated purchases on its platform, setting the stage for a legal battle with wide-ranging implications for the future of online shopping and user autonomy.
The core of the dispute centers on whether AI agents, acting on behalf of a user, should be treated as human shoppers or as unauthorized bots that violate a platform’s terms of service. Amazon alleges that Perplexity’s tool degrades the customer experience and covertly accesses user accounts, while Perplexity frames the move as an anti-competitive tactic by a dominant market player to stifle innovation. The confrontation highlights a growing conflict over who controls a user’s online activity and data as AI agents become more capable of performing complex tasks autonomously. This case could establish a critical precedent for how AI-driven third-party services are allowed to interact with established digital marketplaces.
Violation of Terms and Degraded Experience
Amazon’s primary legal argument is that Perplexity’s Comet AI agent violates the company’s long-standing terms of service, which prohibit the use of automated data mining tools, robots, and other external software that interact with the platform in unsanctioned ways. According to a statement from the e-commerce leader, agentic third-party applications like Comet have an obligation to seek permission before enabling purchasing capabilities to ensure a positive customer experience. The company claims it repeatedly requested that Perplexity remove Amazon from the Comet experience, citing a “significantly degraded shopping and customer service experience” it provides to users.
In a lawsuit filed in the U.S. District Court for the Northern District of California, Amazon accused the AI startup of more serious infractions, including covertly accessing private customer accounts. The complaint alleges that Perplexity intentionally configured its software to disguise the AI agent’s automated activity, making it appear as if a human were browsing the site. This lack of transparency is a key concern for Amazon, which argues that other third-party services that act on behalf of customers, such as food delivery apps ordering from restaurants or online travel agencies booking flights, operate with clear identification. The company also raised concerns that Perplexity’s system poses security risks to customer data and interferes with its ability to provide the tailored shopping environment it has curated over decades.
Accusations of Anti-Competitive Bullying
Perplexity has countered Amazon’s legal actions with sharp criticism, characterizing the demands as “bullying” and a “threat to all internet users.” In a public blog post, the AI startup, which is notably backed by Amazon founder Jeff Bezos, argued that its AI assistant is not a traditional bot but a tool operating directly on behalf of a logged-in user. According to Perplexity, these agents should have the same rights and permissions as the human user they represent, likening Amazon’s stance to an attempt to ban a person from having an assistant or employee act on their behalf.
The startup contends that Amazon’s true motives are not centered on customer experience but on protecting its lucrative business model. Perplexity alleged that the e-commerce giant fears that efficient AI agents will bypass the sponsored results, advertisements, and upsell offers that are integral to its revenue. “Easier shopping means more transactions and happier customers. But Amazon doesn’t care,” the company stated, suggesting the retailer is more interested in influencing purchasing decisions than in facilitating direct, efficient transactions. Perplexity fears that if it is forced to identify Comet as an AI agent, Amazon could use that information to block it and favor its own proprietary shopping bot, known as Rufus.
The Technology of AI Shopping Agents
The conflict revolves around a class of technology known as agentic AI, which can perform multi-step tasks autonomously on the web. Perplexity’s Comet is a browser-based AI assistant that can automate online activities, including searching for specific products and executing purchases without direct human intervention for each step. This distinguishes it from simpler web scrapers or data miners, which typically extract information in bulk. Instead, Comet operates using a user’s credentials, which Perplexity states are stored locally, to perform actions within an active, logged-in session.
Perplexity has drawn a distinction between its AI agents and the bots that Amazon’s policies are designed to prevent, though it has not offered a detailed technical explanation of the legal difference. The core of its argument is that software has traditionally been a tool, like a wrench, controlled by the user. In this view, the AI agent is simply a more advanced tool executing the user’s explicit will. However, the capabilities of these agents to mimic human browsing patterns while operating at machine speed create a gray area that current platform policies and legal frameworks have yet to fully address.
Broader Implications for AI and E-Commerce
This legal showdown is among the first of its kind and is being closely watched across the technology and retail sectors. The outcome could set a powerful precedent for how AI agents are governed and whether they can operate freely on major online platforms. If Amazon prevails, it could embolden other large platforms to erect barriers against third-party AI assistants, potentially stifling innovation and competition. Conversely, a victory for Perplexity could open the doors for a new ecosystem of AI-driven services that offer consumers more direct and efficient ways to interact with online services.
This emerging technology faces headwinds not just from corporate incumbents but also from consumers themselves. An Omnisend survey from February 2025 found that 66% of U.S. consumers were unwilling to let an AI make purchases for them, even if it promised better deals. This consumer hesitancy, combined with legal and security challenges, suggests that the path to widespread adoption of autonomous shopping agents will be complex. The dispute also comes as other platforms update their policies; WhatsApp recently revised its business API terms to ban general-purpose AI chatbots, indicating a growing trend of platforms asserting control over the AI tools that can interact with their services.