Amazon has filed a lawsuit against Perplexity AI, alleging that the startup's browser-based shopping feature covertly accesses customer accounts and disguises AI activity as human browsing. The case highlights growing tensions over the regulation of AI agents—autonomous digital assistants that can perform tasks like ordering products on behalf of users.
Perplexity's browser, Comet, includes an AI agent designed to automate shopping. Amazon argues that this poses unacceptable security risks, citing research by Microsoft showing that AI agents are vulnerable to manipulation during shopping tasks. The lawsuit raises questions about liability and control: whose interests does an AI agent represent, and who is responsible for its misconduct?
Perplexity, valued at $20bn after raising $1.5bn, has faced previous controversies over content scraping and plagiarism. Forbes and Wired have accused the company of copying their work, while The Verge compiled a list of its disputes. Notably, Amazon founder Jeff Bezos has invested in Perplexity twice, adding a layer of irony to the legal battle.
The case is seen as a pivotal moment for the next generation of AI, as it tests the boundaries of autonomous agents and their interaction with major platforms. Meanwhile, AI continues to expand into other domains: AI-generated songs topped Spotify and Billboard charts, and a study estimated that 50,000 AI-generated tracks are uploaded to Deezer daily. In cybersecurity, Anthropic reported detecting a near-fully automated cyberattack by Chinese state-linked hackers using its Claude Code tool.
