This isn’t just another tech lawsuit. This is a glimpse into the future—a messy, high-stakes battle for who, or what, is allowed to shop online.

In one corner, you have Amazon, the $1.9 trillion titan of e-commerce, a walled garden so vast it has its own gravity. In the other, you have Perplexity, the hotshot $20 billion AI startup, armed with a new tool that wants to tear down the walls.

The showdown officially began Tuesday in a San Francisco federal court. Amazon filed a lawsuit accusing Perplexity of “computer fraud.” The weapon in question? An AI tool called Comet, which Amazon claims is an “intruder” that “trespasses” on its site.

Perplexity’s CEO, Aravind Srinivas, has a simpler word for Amazon: “a bully.”

This legal brawl is more than just corporate posturing. It’s the first major shot in the “agent wars,” a fight that will define the next decade of the internet and has massive implications for how we all buy, browse, and exist online.

First, what is an “AI agent”?

To understand the fight, you have to understand the tech.

For the last 20 years, we’ve used the internet for search. You ask Google, “What’s the best vacuum for pet hair?” and it gives you 10 links, 5 video reviews, and a bunch of ads. You, the human, still have to do the work: read the reviews, compare prices, go to Amazon, find the product, add it to your cart, and check out.

An AI agent, or “agentic AI,” is the next step.

An agent doesn’t just answer. It does. The dream is to tell your phone, “Buy me the best-reviewed, sub-$300 vacuum for pet hair and get it here by Friday.” The agent would then do all the work: research, comparison, logging in, and purchasing, all in the background.

Perplexity, like OpenAI and Google, is racing to build these agents. Its “Comet” is a browser agent designed to streamline your online life. Amazon, it turns out, is not a fan.

The lawsuit: “that’s not a human”

According to Amazon’s complaint, this is a story of deception.

The e-commerce giant’s terms of service—that wall of text you scroll past—strictly forbid “any use of data mining, robots, or similar data gathering and extraction tools.“

Amazon claims it’s not against AI; it’s against AI that lies.

The company alleges that Perplexity’s Comet agent logs into a user’s Amazon account and starts making purchases. But to Amazon’s security systems, it doesn’t look like a bot. According to the complaint, Perplexity configured Comet to identify itself as a regular Google Chrome browser user, effectively disguising its automated nature.

“No different than any other intruder,” Amazon argued in its filing, “Perplexity is not allowed to go where it has been expressly told it cannot; that Perplexity’s trespass involves code rather than a lockpick makes it no less unlawful.“

This isn’t their first ’til-the-death. People familiar with the matter said Amazon first asked Perplexity to stop its agents from making purchases back in November 2024. The startup complied.

But then, this August, Perplexity allegedly tried again with its new Comet agent. Amazon detected and blocked it. Perplexity, in a classic tech cat-and-mouse game, reportedly released a new version of Comet to get around the security measure.

That’s when Amazon’s lawyers got involved, sending a cease-and-desist letter on Friday, followed by the full-blown lawsuit on Tuesday.

Lara Hendrickson, an Amazon spokesperson, put it in straightforward terms: “It’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions whether or not to participate.“

Perplexity’s rebuttal: “you’re just scared”

Perplexity isn’t taking this lying down. The startup, led by Aravind Srinivas, came out swinging.

In a fiery blog post, the company called Amazon a “bully” that is “targeting a competitor” and trying to “scare disruptive companies.“

Srinivas’s argument is philosophically profound and, for Amazon, economically terrifying. He argues that an AI agent, when deputized by a user, is the user.

“It’s not Amazon’s job to survey that,” Srinivas said in an interview, arguing that an agent should have “all the same rights and responsibilities” as the human who launched it. In his view, Amazon is trying to distinguish between a human click and an agent’s click as just a tactic to “eliminate user rights.“

Perplexity insists its Comet agent isn’t “scraping” or “training” on Amazon’s data. It’s just taking actions—the same actions a human would take—to make a purchase that its user requested.

“Amazon’s a company that we’ve actually taken a lot of inspiration from,” Srinivas said. “But I don’t think it’s customer-centric to force people to use only their assistant, which may not even be the best shopping assistant.“

That’s the jab. Amazon, of course, is building its own AI agents. It has “Rufus,” an AI assistant to help you browse, and a feature in testing called “Buy For Me,” designed to let you shop on other brand sites from within the Amazon app.

Perplexity is essentially accusing Amazon of wanting a monopoly on AI shopping: You can use bots, but only our bots, inside our garden.

The trillion-dollar subtext: ads

So, why is Amazon really suing?

Sure, “degrading the shopping experience” and “privacy vulnerabilities” are the official reasons. But the existential threat isn’t a bad user experience; it’s a “bad” business-model experience.

Amazon’s most lucrative business isn’t just selling stuff. It’s advertising.

The company makes billions of dollars selling prominent placement on its store. When you search for “running shoes,” the first few results are “Sponsored.” A human shopper sees those ads, might click them, and generate revenue for Amazon.

Now, imagine an AI agent.

You don’t program an agent to find the “top-sponsored shoe.” You program it to find the “best-rated shoe under $100 with free shipping.” The agent, perfectly logical and immune to flashy marketing, would bypass the ads entirely and find the objectively best product based on your criteria.

If AI agents become the primary way to shop, Amazon’s high-margin advertising business could evaporate.

This lawsuit isn’t about computer fraud. It’s about Amazon frantically trying to protect its cash cow from a robot that doesn’t watch commercials.

The awkward dinner party

Here’s the final, ironic twist: Perplexity is a massive customer of Amazon.

While Amazon’s retail side is suing Perplexity, Amazon’s cloud division—Amazon Web Services (AWS)—is happily taking its money.

Srinivas has said his company has made “hundreds of millions” in commitments to AWS. In 2023, AWS even brought Srinivas on stage during its annual trade show, touting Perplexity as a shining example of an AI innovator building its business on Amazon’s infrastructure.

It’s the perfect metaphor for the current tech landscape: The giants are leasing the building blocks to the startups who are openly planning to disrupt them.

This lawsuit is the first of many. It sets the stage for the fundamental question of the next decade: When you go online, who is in charge? The platform, with its terms of service? Or you, with your new AI agent?

The battle of the bots has begun.