9th Circuit Reinstates Perplexity AI Agent on Amazon Platform

Perplexity’s AI shopping agents have received a significant legal reprieve, with a US appeals court overturning an injunction that had previously blocked the startup from operating its tools on Amazon’s platform. The 9th US Circuit Court of Appeals in San Francisco ruled that users, not Perplexity itself, are the entities accessing Amazon through the AI agents, thereby making a violation of federal computer fraud law unlikely to succeed. This decision marks a pivotal moment, representing the first ruling by a US federal appeals court on the legality of AI agents accessing online platforms on behalf of users, a question with broad implications for the burgeoning field of agentic AI.

Key Developments

• A US appeals court has overturned a preliminary injunction that prevented Perplexity’s AI shopping agents from operating on Amazon.
• The court determined that users, not Perplexity, are responsible for accessing Amazon’s platform via the AI agents, weakening Amazon’s claim of computer fraud.
• This ruling is the first by a US federal appeals court to address the legality of AI agents accessing online platforms on behalf of users.
• Amazon expressed disagreement with the decision, stating confidence in its case and evaluating future legal options.
• Perplexity affirmed its commitment to defending internet users’ right to choose their preferred AI tools.

What Happened

Amazon initiated legal action against Perplexity in November, alleging that the startup’s Comet browser and integrated AI agent covertly accessed private Amazon customer accounts. The complaint detailed how Perplexity’s agent could log into user shopping accounts and place orders, raising concerns about security risks and claiming the startup had ignored repeated demands to cease operations. In response, Perplexity characterized the lawsuit as baseless, asserting it was an attempt by Amazon to shield its customers from AI agents that bypass pervasive advertising.

In March, a federal court in California issued a temporary injunction, barring Perplexity from using its agentic AI on Amazon. Judge Maxine Chesney found “strong evidence” that Perplexity accessed password-protected accounts with user permission but without Amazon’s authorization, also ordering the deletion of any copied Amazon data. However, the 9th US Circuit Court of Appeals has now reversed this injunction, concluding that the federal computer fraud law, which prohibits unauthorized computer access to obtain information, was unlikely to apply to Perplexity directly, as the users themselves were deemed to be performing the access.

Why It Matters

This appeals court decision holds significant weight for the evolving landscape of AI and online commerce. It provides initial judicial guidance on the contentious issue of agentic AI systems interacting with established online platforms. Agentic AI, designed to plan, reason, and execute tasks with minimal human oversight, often involves browsing, shopping, and completing transactions across the web. The court’s distinction between a user accessing a platform via an agent and the AI company itself doing so could set a precedent for how future AI-powered tools are regulated and integrated into existing digital ecosystems.

Industry Impact

The ruling has immediate implications for developers of agentic AI and companies operating large online platforms. For AI developers, it offers a degree of clarity and potential validation for models that act on behalf of users, potentially fostering innovation in personal assistants, automated shopping, and web automation. Conversely, for platforms like Amazon, it highlights the challenges in controlling how third-party AI interacts with their services, especially when user consent is involved. Amazon had previously advocated for third-party AI to operate transparently and adhere to established APIs, pointing to its own tools like “Buy For Me” and the “Rufus” assistant. This legal development could necessitate new strategies for platform providers to manage agentic access while balancing user autonomy and platform security.

Analysis

The core of this legal dispute hinges on the attribution of action: who is “accessing” the platform when an AI agent is involved? The appeals court’s reasoning that users themselves access Amazon through Perplexity’s agents, rather than Perplexity directly, is a crucial interpretation of existing computer fraud law in the context of advanced AI. This perspective aligns with Perplexity CEO Aravind Srinivas’s argument that agents acting on behalf of users should possess the same rights as human users. This legal nuance could shape how future regulations differentiate between direct corporate access and user-initiated, agent-mediated interactions.

Amazon’s initial lawsuit underscored concerns about security risks and unauthorized access to private customer accounts. The company’s stance reflects a broader industry challenge: how to maintain control over proprietary platforms and user data when sophisticated AI tools can mimic human interaction. The court’s decision, while overturning the injunction, does not resolve the underlying case, meaning the legal battle over these fundamental questions of AI agency, platform control, and user rights will continue. The ongoing dispute also reveals complex business relationships, with Perplexity being an AWS customer with substantial commitments, and Amazon founder Jeff Bezos among Perplexity’s investors, adding layers to the competitive and legal dynamics.

Future Implications

In the near-term (3–6 months), this ruling is likely to embolden other AI startups developing agentic tools, potentially leading to an increase in new applications designed to automate user tasks across various online platforms. Amazon is expected to pursue its case, potentially refining its arguments regarding security, data privacy, or terms of service violations, rather than solely relying on computer fraud statutes.

Medium-term (1–2 years), the legal precedent set by this case could influence how platforms design their APIs and terms of service to either accommodate or restrict agentic AI. It may also spur legislative discussions around new frameworks specifically tailored to AI agency, as existing laws struggle to keep pace with technological advancements. This could lead to a clearer, albeit potentially more restrictive, regulatory environment for AI agents.

Long-term (3–5 years), the outcome of this and similar cases will significantly shape the competitive landscape of online commerce and personal AI assistants. Depending on final rulings, either platform control could be strengthened, or user-centric AI agents could gain broader acceptance, fundamentally altering how consumers interact with the digital economy and potentially decentralizing some aspects of online shopping and service consumption.

Key Takeaways

• A US appeals court has reversed an injunction against Perplexity, allowing its AI shopping agents back on Amazon.
• The court’s decision hinges on the interpretation that users, not Perplexity, access Amazon through the AI agents.
• This marks the first US federal appeals court ruling on the legality of AI agents accessing platforms on behalf of users.
• Amazon plans to continue its legal challenge, while Perplexity champions user choice in AI tools.
• The underlying legal case remains unresolved, promising further developments in the debate over AI agency and platform control.