Perplexity’s Browser, Comet, Uses AI to Shop for Users
Amazon is taking Perplexity AI to court over its Comet browser. The suit alleges artificial intelligence is used to make purchases on the e-commerce giant’s platform without authorization. The lawsuit further claims the startup’s technology masquerades as human shoppers. This raises serious questions about AI autonomy, online commerce, and data privacy.
What’s Happening & Why This Matters
The Lawsuit
Amazon filed the complaint in the U.S. District Court for the Northern District of California. The commerce retailer is accusing Perplexity AI of violating its terms of service and committing computer fraud. The claim centers on Comet, an AI-powered browser. It can independently log in to Amazon accounts and complete purchases without user input.
According to court documents, Perplexity’s system “covertly accesses customer accounts”. Comet disguises automated actions as legitimate browsing. Amazon says the agentic AI not only bypasses its security systems, it also disrupts the curated experience Amazon built over decades.
“Perplexity’s misconduct must end,” Amazon’s legal team stated. “That its trespass involves code rather than a lockpick makes it no less unlawful.”
Perplexity’s Response
Perplexity fired back in a blog post. Perplexity accuses Amazon of using litigation to bully smaller competitors and stifle innovation. The company defended Comet, insisting that user credentials stay local on the device and never transmit to its servers. It argued that users should be free to choose which AI assistant they use.
“Easier shopping means more transactions and happier customers,” the AI innovator said. “But Amazon doesn’t care — they’re more interested in serving ads and influencing purchases.”
Impact & Reactions
This battle presents the frot discourse between Big Tech and AI startups. They are racing to define how autonomous software interacts with the web. Amazon’s lawsuit targets an emerging category known as agentic AI. These are autonomous systems that act on behalf of humans, including making financial transactions.
The risk is not theoretical. In internal tests, reviewers found Comet capable of purchasing items on Amazon without re-entering login details or payment credentials. That autonomy, Amazon argues, introduces both security vulnerabilities and revenue threats.
Meanwhile, Walmart is taking the opposite route by partnering with OpenAI to allow ChatGPT users to shop its catalog. They are using Instant Checkout. “This is agentic commerce in action,” Walmart announced. The same innovation that Amazon condemns is being embraced elsewhere — by a competitor no less.
Security and retail experts caution that autonomous agents still introduce risks. OpenAI’s security team admitted its own agent, Atlas, sometimes buys incorrect items or acts without confirmation. The fault(s) is due to prompt injection attacks: malicious instructions embedded in websites to manipulate an AI’s behavior.
Perplexity’s response sidesteps the dangers, choosing instead to frame its dispute as a matter of freedom and competition. “Software is becoming labor,” it said. “Amazon doesn’t believe in your right to hire an assistant or an employee acting on your behalf.”
TF Summary: What’s Next
MY FORECAST: The legal clash is another facet of AI-powered commerce. Amazon wants control of its ecosystem; Perplexity offers a future where digital assistants act independently. The courts will decide [in this case] whether that future is innovation — or intrusion. Expect further lawsuits. Eventually, regulation around AI shopping agents will come as governments, corporations, and innovators race to the limits of automation.
— Text-to-Speech (TTS) provided by gspeech