Appeals Court Backs Perplexity in AI Agent Lawsuit Against Amazon

Appeals Court Backs Perplexity’s AI Shopping Agent in Landmark Ruling

The Ninth Circuit Court of Appeals handed Perplexity a significant legal victory Tuesday by overturning a temporary ban on its Comet shopping assistant. This marks the first appellate ruling on whether autonomous AI systems can legally conduct transactions on users’ behalf. According to the court, Amazon is unlikely to succeed in proving Perplexity violated the Computer Fraud and Abuse Act (CFAA), a federal anti-hacking statute at the heart of the dispute.

The conflict between Amazon and Perplexity began in November when Amazon sent Perplexity a cease-and-desist letter, alleging that Comet—a browser tool capable of accessing accounts, making authenticated requests, and executing purchases—had accessed password-protected customer accounts without authorization. Amazon claimed the software disguised its automated traffic as a standard web browser to evade detection, violating both the platform’s terms of service and federal computer fraud law.

A federal judge initially sided with Amazon in March, imposing a preliminary injunction that temporarily blocked Comet from operating on Amazon’s platform. That ruling suggested the company had presented substantial evidence that Perplexity likely committed computer fraud. The Ninth Circuit’s reversal signals a markedly different legal analysis at the appellate level.

A Narrow Reading of Unauthorized Access

The appellate court’s reasoning centered on a critical distinction about who accessed Amazon’s systems. According to the Ninth Circuit, the CFAA requires proving that the defendant—in this case, Perplexity itself—intentionally accessed a computer without authorization. Because Comet operates exclusively with explicit user permission and employs credentials stored on users’ own devices, the court found that users, not Perplexity, are the party performing the access. This interpretation significantly undermines Amazon’s legal theory under federal anti-hacking statutes.

Perplexity’s defense emphasized that Comet functioned only with voluntary user authorization and never exceeded its intended scope. The startup framed the dispute as a consumer autonomy issue, contending that Amazon was attempting to prevent users from deploying AI assistants to manage their own digital activities.

Implications for Agentic AI Systems

The appeals court decision carries implications extending well beyond Perplexity and Amazon. As autonomous artificial intelligence systems become increasingly sophisticated and commonplace, the legal status of AI tools acting on users’ behalf has remained ambiguous. The Ninth Circuit’s framework—distinguishing between a user conducting access through an agent versus an agent conducting unauthorized access itself—provides crucial guidance for how courts may evaluate similar disputes.

Financial services, social media platforms, e-commerce sites, and other online services may soon encounter comparable legal challenges as users deploy AI assistants to automate their interactions. The appellate ruling clarifies that user-authorized AI agents likely do not trigger CFAA liability for the AI provider, provided the agent remains under the user’s direction and control.

Although the underlying lawsuit between Amazon and Perplexity continues, this appellate decision offers courts a framework for evaluating whether autonomous AI systems violate computer fraud law. It could reshape how technology companies approach both AI regulation and consumer tools that automate internet activity. The ruling suggests that AI systems operating with genuine user consent and authorization may occupy legal territory that was previously uncertain.

This decision establishes precedent for how blockchain platforms and crypto services may interpret automated agent behavior, potentially influencing how decentralized systems approach user-authorized autonomous transactions.

Source: Ninth Circuit Court of Appeals, via Decrypt. Not financial advice.