The San Francisco-based 9th U.S. Circuit Court of Appeals has delivered a landmark ruling that fundamentally reshapes how autonomous artificial intelligence systems can operate online. On Tuesday, the appeals court overturned a lower court's temporary ban that had prevented Perplexity, an AI startup, from deploying its agent-based shopping tools on Amazon's platform. The decision represents the first major federal appeals court pronouncement on whether AI agents operating with user authorisation can legally interface with online services, a question with sweeping ramifications for the rapidly expanding field of agentic AI technology.
The dispute originated when Amazon initiated legal proceedings against Perplexity in November, claiming the startup had surreptitiously accessed customer accounts through its Comet browser and associated AI agent. According to Amazon's complaint, Perplexity's system enabled users to grant the AI permission to log into their shopping accounts and complete purchases without direct human intervention for each transaction. Amazon contended that this capability represented both a security vulnerability and a violation of terms of service, and asserted that it had repeatedly requested Perplexity cease the practice. The retail giant further argued that Perplexity's tactics breached the Computer Fraud and Abuse Act, a foundational federal statute governing unauthorised computer access.
Perplexity mounted a vigorous defence, asserting that Amazon's lawsuit lacked legal substance and constituted an anti-competitive "bald attempt" to prevent users from accessing Amazon through AI intermediaries. The company contended that its agents enabled users to filter out the advertising-saturated experience of browsing Amazon directly. This framing shifted the debate from one about corporate security to one about consumer choice and the freedom to employ AI tools for personal shopping assistance.
In March, a federal judge in California initially sided with Amazon, issuing a preliminary injunction that barred Perplexity from operating its agentic tools on the platform. The judge found that Amazon had presented "strong evidence" of a violation of computer fraud laws. However, the appeals court's Tuesday reversal demonstrates a markedly different interpretation of how the law applies to AI agents.
The 9th Circuit's central legal reasoning hinged on a distinction between the AI company itself and the users employing the AI. The court determined that under the relevant statute, it is the users—through Perplexity's AI agents acting on their behalf—who "access" Amazon's platform, not Perplexity as an independent entity. This interpretation proves crucial because it suggests that if a user has legitimate authorisation to access a service, employing an AI agent to do so on their behalf does not constitute unauthorised access. The ruling thereby grants substantial leeway to agentic AI systems operating within the scope of user permissions.
This distinction carries profound implications for the emerging agentic AI sector, which encompasses systems capable of planning, reasoning, and executing tasks with minimal human direction. Companies developing these technologies have long grappled with legal uncertainty about their operational boundaries. The 9th Circuit's ruling suggests that as long as an AI acts under genuine user authorisation, the underlying technology does not itself trigger computer fraud statutes merely by accessing websites on users' behalf.
For Malaysian and Southeast Asian technology observers, the decision warrants close attention. The region's digital economy continues expanding rapidly, with e-commerce platforms, fintech applications, and AI-driven services proliferating. As regional tech companies and international firms develop agentic AI applications for the Malaysian market and beyond, this ruling establishes important precedent about what these tools can legally accomplish. The decision may embolden technology developers to create more sophisticated agent systems while simultaneously raising questions about how such developments will be regulated in jurisdictions with different legal frameworks.
Amazon responded by signalling its determination to continue the legal battle. A company spokesperson stated that Amazon remained "confident in our case" and was "evaluating our next steps," suggesting the retail giant may pursue further appeals or modified legal strategies. This indicates that the dispute between the two companies remains unresolved substantively, with Tuesday's ruling addressing only the preliminary injunction stage.
Perplexity characterised the decision as vindicating user rights. Spokesperson Jesse Dwyer declared the company would "continue to fight for the right of internet users to choose whatever AI they want," framing the outcome as a broader victory for consumer autonomy rather than merely a corporate legal win. This rhetorical positioning attempts to align Perplexity with consumer interests against a more restrictive interpretation of what AI systems can accomplish.
The broader context reveals mounting tensions between technology incumbents protecting their platforms and emerging AI companies seeking to offer new user experiences. Amazon's aggressive legal posture reflects concerns shared across the technology industry about maintaining control over how users interact with their services and what data AI systems can access. Simultaneously, Perplexity's success in this preliminary round suggests courts may be reluctant to impose categorical restrictions on AI technology absent clearer evidence of harm.
The ruling also raises unresolved questions about liability and responsibility. If users authorise AI agents to make purchases on their behalf, who bears responsibility if those purchases prove fraudulent or problematic? The court's framing places legal responsibility with the users rather than the AI company, a position that may prove contentious as agentic systems become more autonomous and users' understanding of their capabilities remains limited.
For regional regulators and policymakers, the 9th Circuit's decision provides a window into how American courts are beginning to conceptualise AI agency and authorisation. As Southeast Asian governments develop their own artificial intelligence governance frameworks, this precedent demonstrates one approach to balancing innovation with consumer protection. However, regional approaches may diverge significantly, particularly regarding data protection and cross-border digital service delivery.
Moving forward, this decision likely accelerates development of agentic AI applications across multiple sectors. Technology companies now have greater confidence that user-authorised AI agents can access online platforms without triggering computer fraud liability. However, the decision does not resolve questions about terms of service violations, data protection compliance, or regulatory requirements specific to financial transactions and consumer privacy. The legal landscape for agentic AI remains unsettled, with the 9th Circuit's ruling addressing one dimension of a far more complex emerging challenge.
