The San Francisco-based 9th Circuit Court of Appeals delivered a significant setback to Amazon on Tuesday, overturning a lower court's temporary injunction that had barred Perplexity from deploying its AI-powered shopping agents on the e-commerce giant's platform. The ruling marks a watershed moment in the rapidly evolving intersection of artificial intelligence and internet commerce, determining that Perplexity's technology did not breach federal computer fraud statutes as Amazon had contended. The decision carries implications far beyond this specific corporate dispute, establishing judicial precedent for how agentic AI systems—autonomous tools capable of browsing, purchasing and transacting across websites—may legally operate within the broader digital ecosystem.
The appellate court's determination hinges on a technical but consequential interpretation of what constitutes unauthorised computer access under US law. Rather than viewing Perplexity itself as the party accessing Amazon's systems, the court reasoned that it is the individual users deploying the AI agent who perform the actual access. This distinction proved decisive in the appeals panel's reasoning. By characterising the transaction as user-initiated rather than company-initiated access, the court sidestepped Amazon's primary legal theory and left the technology largely unfettered to continue its operations. The ruling represents the first time a federal appeals court has tackled this specific question about AI agent access rights, signalling that such determinations will likely shape the regulatory landscape for this nascent technology sector.
Amazon's legal team responded by expressing disappointment with the outcome whilst signalling intent to pursue alternative avenues. The company's statement emphasised that it remained confident in the underlying merits of its case and was actively evaluating next steps, which could include an appeal to the Supreme Court or revised legal strategies. For Amazon, the decision represents a retreat from its original position as the world's dominant e-commerce platform, suggesting that the company's market power alone cannot prevent technological access to its digital properties through judicial means. The setback underscores the challenges major platforms face when seeking to maintain exclusive control over customer interactions and data flows.
Perplexity, the AI startup at the centre of the dispute, hailed the decision as validation of user rights in an increasingly AI-mediated digital landscape. Company spokesperson Jesse Dwyer framed the ruling as a defence against the erosion of consumer choice, asserting that internet users ought to determine which AI tools they employ without interference from dominant platforms. This framing reflects a broader ideological stance within the AI industry that positions agentic tools as liberating technologies that empower users against algorithmic manipulation and platform gatekeeping. Perplexity's rhetorical emphasis on user autonomy appeals to anxieties about corporate control of digital experiences, particularly around the pervasive advertising and data collection practices that characterise contemporary e-commerce.
The original lawsuit, filed by Amazon in November, centred on allegations that Perplexity's Comet browser and associated AI agent were accessing private customer accounts without authorisation, potentially logging into shopping profiles and executing purchases on users' behalf. Amazon contended that this functionality posed genuine security vulnerabilities and that Perplexity had dismissed repeated cease-and-desist communications. The e-commerce company argued that the startup's system essentially bypassed Amazon's terms of service by circumventing its advertising infrastructure—a claim Perplexity dismissed as pretextual. In March, a federal judge in California appeared to accept Amazon's security concerns, issuing the temporary ban after finding strong evidence that Perplexity's methods violated the Computer Fraud and Abuse Act.
Perplexity's defence characterised the lawsuit as a commercially motivated attempt to prevent consumers from using advanced AI tools that reduce their exposure to Amazon's advertising ecosystem. The startup's counter-argument emphasised that its agents operated transparently on behalf of users who had explicitly authorised their deployment, thereby distinguishing its conduct from surreptitious hacking. This reframing proved persuasive to the appellate panel, which adopted Perplexity's interpretation of existing statute and determined that the relevant law's language contemplates user-directed rather than company-directed access. The legal logic underlying the court's reasoning may extend to other AI agents seeking to interact with major platforms, potentially opening pathways for competing technologies to access and navigate protected digital spaces.
Agentic AI systems represent a frontier in machine learning applications, distinguished by their capacity to plan, reason through problems and execute multistep tasks with minimal human intervention. Unlike traditional AI interfaces that respond to direct queries, these agents can autonomously navigate websites, synthesise information from multiple sources and take actions in response to user-defined objectives. Retailers, financial services providers and other data-sensitive industries view such capabilities with particular concern, fearing both security implications and loss of direct customer relationships. The appeals court's decision suggests that legal frameworks developed in the pre-AI era may inadequately address these new technological realities, potentially requiring legislative rather than purely judicial solutions.
For Malaysian and Southeast Asian technology companies and platforms, the ruling carries significant implications as the region develops its own AI and e-commerce sectors. The precedent established by the 9th Circuit may influence how regional courts approach similar disputes, particularly as local startups develop AI tools that interact with cross-border digital services. The decision suggests that platform access restrictions based on traditional computer fraud statutes face substantial legal headwinds in jurisdictions with similar statutory frameworks. Additionally, the ruling may encourage greater investment in agentic AI development across Asia-Pacific, as companies gain confidence that such tools can legally interface with major online services.
The broader regulatory environment surrounding agentic AI remains unsettled, with this court decision representing only one data point in an emerging legal landscape. While the 9th Circuit favoured Perplexity's interpretation, other courts could reach different conclusions, and legislative bodies may ultimately impose restrictions that judicial decisions have declined to impose. Amazon has not exhausted its legal remedies, and the company's potential appeal could generate further appellate guidance. Simultaneously, policymakers at federal and state levels may conclude that this technology requires more comprehensive regulation than existing computer fraud statutes can provide, particularly given legitimate concerns about security, privacy and platform manipulation.
The decision ultimately reflects an ideological choice about how digital markets should function and who holds power within them. By declining to read computer fraud statutes expansively against AI agents, the court privileged user autonomy over platform control, at least in this preliminary stage of litigation. This stance aligns with longstanding legal preferences for technology-neutral rules that neither favour nor discriminate against particular innovations. However, the same approach may prove inadequate for addressing genuine security risks or protecting consumer interests if agentic systems become vectors for fraud, identity theft or account takeover schemes. As these technologies mature and proliferate, the tension between enabling innovation and protecting digital security will likely intensify, potentially requiring more sophisticated legal frameworks than currently exist.
