Perplexity AI Hit With Privacy Lawsuit While Appeals Court Greenlights Its Bots

Perplexity AI is facing a proposed class‑action lawsuit alleging it violated California privacy laws by secretly sharing users' personal information and AI chat conversations with Meta and Google. The suit, filed in the US District Court for the Northern District of California (case 3:26‑cv‑02803), names all three companies as defendants, according to Insurance Journal.

The plaintiff, a Utah man identified as "John Doe," claims he shared highly sensitive information with Perplexity — family finances, tax obligations, and investment strategies — without knowing the data was being transmitted to third parties. According to the complaint, trackers are downloaded to user devices immediately upon logging into Perplexity's homepage, giving Meta and Google full access to conversations between users and the AI search engine.

The suit alleges Perplexity embedded "undetectable" tracking software into its search engine code that automatically transmits user conversations to third parties. Critically, user data is shared even when "Incognito" mode is active. Meta and Google are accused of exploiting this data for ad targeting and reselling it to additional third parties, in violation of federal and state computer privacy and fraud laws.

Perplexity Chief Communications Officer Jesse Dwyer told:¹ "We have not been served any lawsuit that matches this description so we are unable to verify its existence or claims." Meta pointed to a Facebook help page stating it is against the company's rules for advertisers to send sensitive information.

While the privacy lawsuit is new, Perplexity has been fighting a separate legal battle with Amazon since November 2025 over its Comet AI shopping agent — a tool that lets users ask an AI assistant to find items on Amazon and make purchases on their behalf.

Amazon sued Perplexity in November 2025, alleging Comet accessed password‑protected portions of Amazon customer accounts without Amazon's authorization, disguised automated activity as human browsing, and ignored repeated demands to stop, as reported by CNBC.

On March 9, 2026, US District Judge Maxine Chesney granted Amazon's request for a preliminary injunction. Her key finding: Perplexity accessed accounts "with the Amazon user's permission but without authorization by Amazon," per.² The ruling relied on the Facebook v. Power Ventures precedent, which holds that a cease‑and‑desist letter can terminate access rights that might otherwise exist.

The injunction required Perplexity to prohibit Comet from accessing Amazon accounts and delete any collected customer data. But Perplexity appealed immediately, arguing the ban would cause "devastating harm" to its business, according to CyberScoop.

On March 30, 2026, the Ninth US Circuit Court of Appeals granted Perplexity an administrative stay, pausing the lower‑court injunction while the appeal is considered, according to CyberScoop. Perplexity filed its formal appeal on April 1, and Amazon's response is due April 22.

The appeals court's decision to grant the stay is significant — it means the court saw enough merit in Perplexity's arguments to keep Comet operating while the case proceeds. Perplexity's core argument: a Comet user accessing Amazon from her own computer is no different from a Safari user doing the same. "AI agents don't have eyeballs to see the pervasive advertising Amazon bombards its users with," Perplexity argued, per.⁴

Amazon's counterargument focuses on its advertising business, which generated roughly $68.6 billion in ad revenue in 2025. AI‑generated traffic must be detected and filtered before advertisers can be billed for legitimate human impressions. Amazon argued that filtering Comet's automated traffic required building new detection systems, and that employees worked "numerous hours" on the problem, per.²

Both cases — the privacy lawsuit and the Amazon dispute — converge on a single question that will define the next era of AI: who controls the relationship between an AI agent and the services it interacts with?

The Amazon case asks whether user permission is sufficient for an AI agent to act on a third‑party platform, or whether the platform must also authorize the interaction. If the Ninth Circuit upholds Judge Chesney's ruling, platforms would gain broad power to block AI agents regardless of what users want. If it overturns the ruling, AI agents could operate freely on any site as long as their users consent.

The financial stakes are enormous. Amazon has blocked dozens of outside AI agents, including OpenAI's ChatGPT, while building its own assistant, Rufus, which drove nearly $12 billion in incremental annualized sales in 2025, per.⁴ Walmart and Target are taking the opposite approach — testing ways to work with third‑party AI shopping tools while maintaining their own role in the transaction.

Consumer behavior is shifting fast. According to PYMNTS Intelligence, 70% of consumers are open to using AI agents for shopping, and over 50% of consumers who already rely primarily on AI platforms prefer to complete purchases inside those environments rather than being handed off to a retailer's site.

Both Perplexity cases carry direct implications for anyone building AI agents, chatbots, or tools that interact with third‑party platforms.

Privacy‑by‑design is now a legal requirement, not a nice‑to‑have. The Perplexity lawsuit alleges that user data was shared through hidden trackers — including in Incognito mode. If you're building an AI tool that processes user conversations, assume that any data‑sharing mechanism will be scrutinized in court. Document every data flow and get explicit consent.

The CFAA is being weaponized against AI agents. Amazon's case against Perplexity relies on the Computer Fraud and Abuse Act — the same law used against hackers. The argument is that an AI agent accessing a site without the platform's explicit authorization is equivalent to unauthorized access. If you're building agents that interact with third‑party services, understand that each service's terms of service may become a legal minefield, regardless of what your users consent to.

Platform vs. user consent is the defining legal question. The Ninth Circuit's eventual ruling will set a precedent that reaches far beyond Amazon and Perplexity. If platforms win, every AI agent builder will need explicit API partnerships with the services they interact with. If agents win, user consent alone could be enough — but expect platforms to deploy increasingly aggressive technical countermeasures.

Track the Facebook v. Power Ventures precedent. Judge Chesney's injunction relied on this 2016 Ninth Circuit ruling, which held that a cease‑and‑desist letter can terminate access rights. This precedent is the strongest legal tool platforms have against AI agents. Builders should assume any platform can issue a cease‑and‑desist at any time and have a contingency plan.