The DoorDash Problem: How AI Browsers Are a Threat to Amazon (www.theverge.com)

Amazon has opened a high-stakes legal front in what’s being called the “DoorDash problem”: the risk that AI browsers and agentic assistants will interpose themselves between consumers and service providers, stripping away the ads, reviews, loyalty programs and upsells that companies use to monetize customers. Amazon sued Perplexity after its Comet browser — an AI-first Chromium-based browser that combines Perplexity’s and third-party models — built a shopping agent that automated purchases on Amazon.com. Amazon alleges Perplexity ignored take-down requests, masked agents as human Chrome users, and bypassed technical blocks, and it has invoked the Computer Fraud and Abuse Act; Perplexity calls the suit bullying. For the AI/ML community this crystallizes both a technical and economic battleground: agentic systems can reduce websites to commoditized databases if they routinely transact on users’ behalf, undermining platform economics and changing incentives for data, UI, and APIs. Key technical points include agent automation of browsing and buying, circumvention of bot-detection and authentication, and the rise of AI-enabled browsers (Comet, Copilot-in-Edge, Atlas) that reshape how models interact with the web. The outcome could force new standards—agent-aware APIs, authentication protocols, and legal frameworks—or provoke defensive measures by platforms and regulators. For researchers and product teams, this signals urgent work on safe agent behavior, traceable automation, and cooperative interfaces that preserve service providers’ ability to monetize while enabling agent convenience.