Amazon vs. Perplexity AI: Why Amazon Is Trying to Block Perplexity’s AI Agent from Its Store

Introduction: A New Battleground in the AI Race

A heated conflict has broken out between Amazon, the e-commerce titan, and Perplexity AI, a rising star in the generative AI space. At the heart of the dispute lies Comet, Perplexity’s agentic-AI browser, which can autonomously shop on behalf of users — including placing orders on Amazon itself. In November 2025, Amazon escalated the matter by sending a cease-and-desist letter, demanding Perplexity disable Comet’s shopping capability on its platform.

This conflict illustrates how AI agents are challenging traditional business models, raising questions about user autonomy, platform control, and the future of commerce. Below, we unpack this clash from all major angles: Amazon’s claims, Perplexity’s defense, the wider significance, and what it may mean for the future of AI-powered shopping.

What Is the Dispute About?

Amazon’s Allegations: Violating Terms & Degrading Experience

Amazon argues that Perplexity’s Comet assistant is violating its Terms of Service by acting on behalf of users without clearly identifying as an AI agent. According to Amazon:

1. Lack of transparency: Comet is accused of failing to disclose that it is an automated agent, instead presenting itself as a regular browser or user.
   
2. Degraded shopping experience: Amazon claims the agent yields a worse experience — for example, by bypassing personalization, misreporting delivery estimates, or not honoring typical Amazon features.
   
3. Privacy & account security: There are concerns about how Comet logs into users’ Amazon accounts, possibly posing risks to data security.
   
4. Computer fraud: Amazon accuses Perplexity of “computer fraud” because agents are not properly identifying themselves and may be masquerading as human browsers.
   

In its statement, Amazon said third-party applications making purchases should “operate openly” and must “respect service provider decisions whether or not to participate.”

Perplexity’s Defense: Autonomy, User Choice & Innovation

Perplexity pushes back hard against Amazon’s claims:

• It argues that its Comet Assistant acts on behalf of real users, using their credentials, so it should enjoy the same permissions as a human user.
  
• According to Perplexity, credentials remain stored locally on the user’s device — not on Perplexity’s servers — implying that it is not directly controlling or storing sensitive data.
  
• In a blog post titled “Bullying Is Not Innovation,” Perplexity frames Amazon’s move as an attempt to stifle competition and maintain its ad-driven business model.
  
• It claims that Amazon would prefer users to rely on its own AI assistant (e.g., Rufus) because third-party agents could bypass sponsored listings, upsells, and confusing offers that benefit Amazon’s advertising model.
  

Broader Implications: Why This Matters for AI, Commerce & Regulation

1. The Rise of Agentic AI Agents

Perplexity’s Comet is part of a growing class of agentic browsers — agents empowered by AI to navigate, act, and transact on behalf of users. As agents become more autonomous, tensions emerge over how much control platforms (like Amazon) want to retain and how much autonomy users intend to give their chosen agents. This dispute marks one of the first major tests in determining who will strike that balance.

2. Platform Control vs. User Choice

Amazon’s demand that Perplexity identify itself as an agent could force a precedent: only “approved” agents or in-house bots may gain full access. This raises concerns about gatekeeping, where platforms might restrict which AI assistants can act on their site, potentially limiting user freedom.

From Amazon’s perspective, unauthorized agents could degrade the shopping experience and undermine its business model. From Perplexity’s angle, prohibiting third-party agents is anti-competitive and undermines innovation.

3. Advertising and Monetization

A key part of Perplexity’s argument is economic: if AI agents do the shopping, they might skip sponsored links, ads, and upsells — Amazon’s most profitable real estate. Allowing these agents could erode Amazon’s advertising revenue model. By pushing back, Amazon may be defending its bottom line, not just user experience.

4. Legal & Regulatory Precedent

Amazon’s legal move could set precedents for how AI agents are permitted (or restricted) to interact with web platforms. As more AI assistants emerge — not just for shopping, but booking, data lookup, or web navigation — regulators and businesses will need to define rules of engagement for autonomous agents:

• Do agents need to explicitly identify themselves?
  
• To what extent can they operate on behalf of a user?
  
• What obligations do they have to the platforms they interact with?
  

This case could become a cornerstone for future policy and legal battles in the AI economy.

Real-World Stakes: What’s at Risk for Both Sides

For Perplexity

• Innovation at risk: If forced to disable Comet’s shopping capabilities, Perplexity may lose a major differentiator.
  
• User trust: Perplexity argues that users should be free to choose their own AI assistants and not be blocked from using their tools.
  
• Business model tension: The company positions itself not just as a search tool, but as a shopping assistant. Curtailing that could force it to reinvent parts of its strategy.
  

For Amazon

• Loss of ecosystem control: Agents like Comet could redirect how users shop, bypassing Amazon’s optimized UI and recommendation engine.
  
• Ad revenue loss: As mentioned, if agents skip sponsored products, Amazon’s ad business could take a hit.
  
• Security concerns: Allowing third-party agents to act with full user permissions carries risk, especially if the agents misbehave or are exploited.
  

Technical, Ethical, and Security Dimensions

Technical Risks

• Session handling: Amazon claims Comet misrepresents itself as a generic browser (like Chrome), which could violate its terms.
  
• Credential usage: While Perplexity says credentials remain local, the flow of using them to make purchases is delicate and must be securely handled.
  
• Bot detection: Platform-level defenses may treat Comet agents as bots, suitable to blocking or restrictions.
  

Ethical Considerations

• Transparency: Should agents be required to clearly identify themselves when acting on behalf of a user?
  
• Fairness: How do we ensure that agentic tools don’t undermine smaller sellers or sidestep platform rules?
  
• User consent: Users need to understand when their actions are being automated and what that means for their data.
  

Security Risks

• Credential exposure: Even if stored locally, automated usage increases the risk that credentials are misused or compromised.
  
• Fraud risk: If agents act incorrectly, they could make unintended or fraudulent purchases.
  
• Manipulation risk: As with other AI systems, malicious actors could exploit agents via hidden prompts or UI vulnerabilities. (Note: in other agentic browser contexts, security researchers have demonstrated such risks.)
  

Context: The Emerging AI Agent Wars

This isn’t just about Amazon vs. Perplexity — it reflects a broader trend:

• AI agents everywhere: Tech companies and startups are racing to build agents that automate more of our digital lives — shopping, booking, researching.
  
• Platform pushback: Large platforms (retail, travel, social) may push back on agents that operate independently, especially when their business models rely on ads, upsells, or curated experiences.
  
• Legal & regulatory fog: Currently, there is no settled legal framework for how fully autonomous AI agents should interact with web services, making disputes like this a proving ground.
  

What Happens Next: Potential Scenarios

Perplexity Concedes or Modifies

• It could require Comet to explicitly identify as an AI agent when making orders.
  
• It might limit Comet’s Amazon-shopping feature to comply with Amazon’s terms.
  
• Alternatively, Perplexity could negotiate a partnership with Amazon — for example, integration with Amazon’s own preferred agents.
  

Legal Battle

• Amazon may escalate to a lawsuit if Perplexity refuses to comply. In fact, legal action has already been reported.
  
• Courts could define important new rules for agent-platform interaction, potentially requiring agent identification or authentication mechanisms.
  

Regulation Emerges

• Regulators may step in to mandate transparency for AI agents, requiring them to identify themselves when performing actions.
  
• Policy frameworks could emerge around consumer rights, data usage, and agent accountability.
  

Ecosystem Fragmentation

• Platforms may increasingly close off parts of their services to third-party agents to preserve control.
  
• Only a few “approved” agents might remain, suitable to a more constrained AI ecosystem.

Conclusion: A Defining Moment for Agentic AI

The showdown between Amazon and Perplexity over Comet’s shopping agent is more than just a legal spat — it’s a defining moment in how AI agents will operate on the open web. On one side, Amazon is defending its curated, ad-driven ecosystem; on the other, Perplexity argues for user autonomy and innovation.

As AI continues to evolve, these tensions will likely intensify. Will platforms decide to shut out third-party agents or regulate them strictly — or will a cooperative, open model of AI agency win out? The legal and technical precedents set here could shape the future of consumer-facing AI for years to come.

FAQs

Q: What exactly did Amazon demand in the cease-and-desist?

Amazon demanded that Perplexity block its Comet AI assistant from making purchases on Amazon’s website, accusing it of violating terms of service and not clearly identifying itself as an AI agent.

Q: Why does Amazon say Comet provides a “degraded shopping experience”?

Amazon claims that when Comet acts on behalf of users, it bypasses personalization, miscalculates delivery times, and doesn’t use Amazon’s optimized UI — harming the user experience.

Q: How does Perplexity defend its actions?

Perplexity says that Comet acts with user permission, using credentials stored on the user’s device, so it functions like a user, not a bot. They argue Amazon’s efforts to block them restrict user choice.

Q: Is this about ad revenue?

Partly. Perplexity argues that Amazon’s real concern is protecting its advertising business, because third-party agents may skip ads, sponsored search results, and upsells.

Q: Could this set a precedent for other AI agents?

Yes. The outcome may influence how future AI assistants (shopping, booking, navigation) are allowed to interact with large platforms — whether they must identify themselves, or be restricted or approved.