The San Francisco-based 9th Circuit Court of Appeals has dealt a significant setback to Amazon in its legal battle against Perplexity, overturning a temporary ban that had prohibited the AI startup from deploying its shopping tools on the e-commerce giant's platform. The Tuesday ruling represents a watershed moment in the evolving landscape of artificial intelligence law, as it constitutes the first appellate decision to directly address whether AI agents acting on behalf of users can legally access online platforms without explicit permission.
The dispute centres on fundamental questions about how emerging agentic AI systems—programmes capable of planning, reasoning, and executing tasks with minimal human oversight—should operate within the digital ecosystem. Amazon had claimed that Perplexity violated federal computer-hacking statutes by enabling its Comet browser and associated AI agent to access private customer accounts, log in with user credentials, and place orders without direct human intervention at each step. The retail giant characterised the practice as a security threat and argued that Perplexity had disregarded multiple cease-and-desist requests.
Perplexity's response to the lawsuit has centred on a distinction that the appeals court ultimately found persuasive: the company contends that its users, rather than Perplexity itself, are the ones accessing Amazon's platform when they deploy the AI agent. This technical distinction proved decisive in the court's analysis of the Computer Fraud and Abuse Act, the federal statute that Amazon had invoked. The appeals panel concluded that Perplexity had not itself violated the law's prohibition against accessing computers or obtaining information without authorisation, since the actual access occurred through users' deliberate actions and credentials.
Amazon's legal position had initially gained traction in March when a federal judge in California granted a temporary ban on Perplexity's shopping functionality, finding that the tech giant had presented compelling evidence of wrongdoing. However, the higher court's reversal suggests that the lower court's reasoning—that Perplexity bore direct responsibility for the platform access—does not withstand scrutiny under appellate standards. The decision has implications far beyond the immediate dispute between these two companies, as it establishes a legal framework that could permit a wide range of AI-powered tools to autonomously interact with web platforms on users' behalf.
The ruling carries substantial significance for the rapidly expanding field of agentic AI applications. As these systems become increasingly capable of browsing, purchasing, booking, and executing transactions across the internet, companies deploying such tools have gained judicial clarity that users' authorisation may be sufficient legal protection, even when the underlying technology operates with substantial autonomy. This interpretation could accelerate development and deployment of more sophisticated AI agents designed to handle routine digital tasks.
Amazon responded to the decision with measured disappointment, issuing a statement emphasising that the company remains confident in its underlying case and is evaluating subsequent legal strategies. The company has not ruled out appealing further or pursuing alternative legal angles. An Amazon spokesperson stated the company respectfully disagrees with the preliminary injunction decision and continues to believe its claims have merit. The firm's frustration likely stems from the practical implications: the ruling allows Perplexity's AI to resume accessing Amazon's shopping functionality, potentially frustrating Amazon's control over its platform and the data flowing through it.
Perplexity, conversely, hailed the decision as validation of user rights in the AI era. Company spokesperson Jesse Dwyer released a statement emphasising that the firm will continue fighting for internet users' freedom to select their preferred AI tools. Perplexity's framing of the dispute as fundamentally about user choice reflects a broader narrative within the AI industry that attempts to position aggressive platform access as consumer empowerment. The company had previously characterised Amazon's objections as a transparent attempt to block AI agents that cannot display the advertising pervasive on Amazon's website, suggesting that the retailer's real motivation was commercial rather than security-based.
For Malaysian and Southeast Asian stakeholders, this ruling carries tangential but meaningful consequences. As AI agents become normalised for shopping and financial transactions, regional e-commerce platforms and technology companies must anticipate similar legal and technical challenges. The decision establishes that providing user credentials or consent may constitute a complete defence against claims of unauthorised access, a principle that could reshape how regional platforms approach API access and third-party integration policies. Companies operating across multiple jurisdictions must now grapple with the question of whether US appellate precedent will influence courts elsewhere in Asia-Pacific.
The broader legal landscape surrounding AI regulation remains unsettled across the region. While the US court has sided with permissive access principles, other regulatory regimes—particularly those emphasising data protection and platform sovereignty—may reach different conclusions about similar practices. The European Union's Digital Markets Act and emerging AI regulations in various jurisdictions suggest that not all regions will adopt the US approach of prioritising user autonomy over platform protection. Southeast Asian regulators and policymakers should monitor this space closely as they develop their own frameworks for AI accountability.
The case also highlights tensions between competing interests in the digital economy. Amazon's position reflects concerns about platform security, data protection, and the concentration of intermediary power in the hands of third-party AI systems. Perplexity's victory reflects the counterargument that restrictions on AI capabilities ultimately limit consumer choice and innovation. This philosophical divide will likely define regulatory battles over AI agents in coming years, with courts and legislatures worldwide grappling with where to draw the line between protecting platform integrity and enabling technological progress.
Looking ahead, the 9th Circuit's decision may prompt legislative responses. Amazon and other platforms might seek statutory protections that explicitly prohibit certain forms of unauthorised automated access, even when mediated through user authorisation. Alternatively, Congress could establish clearer standards for what constitutes valid user consent in the context of AI agency. Until such developments occur, the practical effect of this ruling is to permit a significant expansion of AI agent capabilities and to establish a legal footing for companies worldwide to build more sophisticated autonomous systems that interact with existing digital infrastructure.
