Artificial intelligence is no longer just a tool consumers and businesses use to draft emails or generate images. A new generation of AI systems, known as agentic AI, can take autonomous actions on a user’s behalf: browsing the web, placing orders, navigating platforms, and interacting with third-party systems without a human in the loop at every step. That capability is creating a new category of legal exposure for ecommerce platforms and brands, and a federal lawsuit filed by Amazon against Perplexity AI offers one of the earliest and most closely watched examples of what that exposure looks like.
The Dispute
Amazon filed suit against Perplexity AI in federal court in California’s Northern District in November 2025, asserting violations of the Computer Fraud and Abuse Act (CFAA) and California’s Comprehensive Computer Data Access and Fraud Act. At the center of the case is Perplexity’s AI-powered browser, called Comet, which Amazon alleges was used to access customer accounts on Amazon’s platform with user permission but without authorization from Amazon.
According to Amazon’s complaint, Comet’s agentic AI enables users to automate tasks like browsing, shopping, and placing orders, but does so by misrepresenting itself as Google Chrome rather than disclosing that it is an AI agent. Amazon maintains this conduct violates its Conditions of Use, which have been updated to address automated and AI-driven activity, including requiring transparent identification of AI agents and prohibit concealment of automated activity. The complaint describes the conduct as a digital trespass: Amazon’s terms tell automated agents where they cannot go, and Perplexity went there anyway by disguising itself as something it was not.
The CFAA claims rest on the allegation that Perplexity knowingly accessed protected computers without authorization, obtaining user account information and transmitting it to its own servers, causing more than $5,000 in damages, which is the statutory threshold for a CFAA civil claim. The California state law claims add further exposure for unauthorized access, copying, and alteration of data. Amazon sought a preliminary injunction to halt further unauthorized access, arguing that the ongoing harm to its systems, customer trust, and reputation was irreparable. In March 2026, the district court granted Amazon’s motion for a preliminary injunction, finding a likelihood of success on the merits, although that order has been stayed pending appeal to the Ninth Circuit.
What Is Agentic AI and Why Does It Matter for Ecommerce?
Agentic AI refers to systems that can pursue a goal autonomously by breaking it into steps, selecting tools, and executing tasks, rather than simply responding to prompts. Systems like Perplexity’s Comet, and enterprise agents from companies like Microsoft and Amazon itself, are designed to take real-world actions inside business and commercial systems on behalf of users.
For ecommerce brands, this creates a genuinely new set of questions. An AI agent that can browse an Amazon store, read product listings, access pricing data, monitor inventory, or simulate purchase behavior can do all of those things at a scale and speed that no human user could match. When those agents are operated by a competitor, a data broker, or a bad actor, the implications for pricing strategy, competitive intelligence, and customer data security are significant.
The Amazon-Perplexity case is also notable for what it reveals about the limits of platform terms of use as an enforcement mechanism. Amazon’s Conditions of Use expressly address AI-driven or automated activity and require transparent identification. That specificity makes the CFAA argument stronger because courts have sometimes struggled with CFAA claims where the “authorization” question is murky, but a platform that expressly addresses agent activity and prohibits misrepresentation has a cleaner record. Brands and platforms that have not yet revisited their terms of use in light of agentic AI are operating with a gap that could prove costly.
Practical Implications for Brands
The Amazon-Perplexity litigation raises three practical questions that ecommerce brands should be asking about their own operations.
The first is whether their platform agreements and terms of use address AI agents specifically. Generic language prohibiting automated access or scraping may not be sufficient to support a CFAA claim against an agentic system that argues it was acting with user authorization. Terms should expressly address agentic AI, require disclosure of automated activity, and prohibit misrepresentation of agent identity.
The second is cybersecurity. Amazon’s complaint describes Comet as having created potential security risks for customer data stored in connected accounts. Any agentic tool that interacts with a customer’s account on a third-party platform, including an ecommerce storefront, is a potential vector for credential exposure, data leakage, or account takeover. Brands that offer customer accounts, loyalty programs, or stored payment data should be evaluating whether their security architecture is hardened against agentic access, not just traditional human-driven account compromise.
The third is the competitive intelligence dimension. AI agents can observe, record, and analyze pricing, inventory, and customer behavior data at a scale that creates real asymmetries between brands that deploy these tools and those that don’t and raises questions about whether that activity crosses the line from competitive intelligence into misappropriation or unfair competition. That line is not yet clearly drawn in the law, but the Amazon lawsuit signals that major platforms are prepared to draw it themselves.
The Bigger Picture
Perplexity has framed Amazon’s lawsuit as an “exclusionary tactic” designed to protect Amazon’s own AI initiatives from competition. That framing is worth noting, courts and regulators will likely be asked to distinguish between legitimate platform security enforcement and anticompetitive gatekeeping as agentic AI becomes more prevalent. For most brand owners, however, the immediate lesson from this litigation is simpler: the legal infrastructure governing who can access your digital storefront, and on what terms, needs to keep pace with a rapidly changing technological environment.
Contact
For questions about ecommerce terms of use, platform security, and the legal implications of AI in online commerce, please contact KJK eCommerce attorney Kyle Stroup (KDS@kjk.com).