On 4 August 2026, the US Court of Appeals for the Ninth Circuit vacated a preliminary injunction that had prevented Perplexity's Comet AI shopping agent from accessing Amazon. The court found that Amazon was unlikely to succeed on its Computer Fraud and Abuse Act claim because the users — not Perplexity — were the parties accessing Amazon through the agent. The full opinion is publicly available from the Ninth Circuit.
_The underlying lawsuit remains active, and Amazon said it is evaluating its next steps. But the ruling is the first federal appellate decision addressing AI agents acting on users' behalf under the CFAA — and its commercial implications extend well beyond the two companies involved.
What the Court Actually Decided
Amazon sued Perplexity in late 2025 after sending a cease-and-desist letter demanding that Comet stop accessing Amazon's platform. A district court granted Amazon a preliminary injunction in March 2026. The Ninth Circuit stayed that order while it reviewed the case, then vacated it entirely on 4 August.
The court's reasoning was precise: under the CFAA, the relevant question is whether the party accessing the computer system had authorisation. Because users authorised Perplexity's agent to act on their behalf, the court found that Amazon was unlikely to prove the agent itself was an unauthorised accessor. The ruling does not resolve the full merits of Amazon's case, and it does not establish an unrestricted right for every AI agent to access every platform.
What it does establish is a significant precedent: a user's delegation to an AI agent may be legally equivalent to the user's own access, at least at the preliminary injunction stage under the CFAA.
The Deeper Question: Who Owns the Buying Journey?
Amazon's commercial model depends partly on controlling the interface through which products are ranked, promoted and purchased. Its advertising business — which generated over $56 billion in revenue in 2024 — relies on sponsored placements, merchandising algorithms and the discovery layer that sits between a customer's intent and a completed purchase.
Agentic commerce may unbundle that interface. When a customer delegates their buying intent to an external agent, the agent may:
- bypass sponsored product placements entirely;
- compare products based on structured data rather than visual merchandising;
- select based on price, availability and policy — not advertising investment;
- complete the transaction without the customer ever seeing Amazon's UI.
The platform becomes fulfilment and inventory infrastructure. The agent becomes the customer's interface. Retail advertising loses some influence over the final decision.
This is not a hypothetical future state. Perplexity's Comet is already doing this for users who choose to use it. The court has now said that doing so is not obviously illegal.
What This Means for Brands and Marketers
The ruling has three immediate implications for brands selling through Amazon and other platforms:
1. Structured product data becomes a competitive advantage. Agents evaluate products based on machine-readable signals: pricing accuracy, availability, specifications, reviews and policy clarity. Brands with incomplete or inconsistent product feeds are invisible to agents that rely on structured data rather than visual merchandising.
2. Advertising investment may not translate to agent visibility. Sponsored placements are a human-facing signal. An agent optimising for a user's stated criteria — lowest price for a specific specification, fastest delivery, best return policy — may ignore paid placements entirely. Brands need to win on the underlying product attributes, not only on advertising spend.
3. Platform access terms are under pressure. If courts continue to find that user-delegated agent access is equivalent to user access, platforms will need to revise their terms of service, API strategies and authentication flows to address agentic intermediaries explicitly. Brands that depend on platform exclusivity or proprietary data advantages may face new competitive dynamics.
The Agentic Commerce Readiness Framework
At Integrated.Social, we define Agentic Commerce Readiness across seven dimensions:
- Structured product data: complete, accurate, machine-readable feeds with real-time pricing and availability;
- Policy clarity: return, warranty, fulfilment and authentication policies stated in agent-parseable formats;
- Evidence quality: verified reviews, specifications and third-party validation that agents can evaluate;
- Transaction access: API and authentication flows that support delegated access without requiring human UI interaction;
- Agent-facing optimisation: content structured for AI evaluation, not only human persuasion;
- Attribution architecture: tracking that captures agent-intermediated journeys, not only direct sessions;
- Governance: clear policies on which agents may access which systems and under what conditions.
Brands that invest in these dimensions now are building the infrastructure for a commerce environment where agents, not websites, determine product consideration and purchase. Those that wait for the legal landscape to fully settle may find that the structural advantages have already accrued to competitors who moved earlier.
The Broader Pattern
The Perplexity ruling is one data point in a larger pattern. The Rezolve AI revenue evidence shows that agentic commerce is already generating real commercial outcomes. The OpenAI enterprise trust debate shows that agent governance is becoming a board-level concern. And the AEO and AI search visibility work we do with clients shows that the brands being cited and selected by AI systems are those that have invested in structured, evidence-led content — not those relying on legacy advertising infrastructure.
The court has not resolved who owns the buying journey. But it has made clear that the answer will not be determined by platform terms of service alone. It will be determined by which brands, agents and platforms build the most useful, trustworthy and accessible infrastructure for delegated commerce.
Modi Elnadi is the founder of Integrated.Social, a B2B AI marketing agency specialising in agentic AI lead generation, AEO/GEO and performance marketing. He has been working at the intersection of AI and commercial marketing since 2014.
On 4 August 2026, the US Court of Appeals for the Ninth Circuit vacated a preliminary injunction that had prevented Perplexity's Comet AI shopping agent from accessing Amazon. The court found that Amazon was unlikely to succeed on its Computer Fraud and Abuse Act claim because the users — not Perplexity — were the parties accessing Amazon through the agent. The full opinion is publicly available from the Ninth Circuit.
_The underlying lawsuit remains active, and Amazon said it is evaluating its next steps. But the ruling is the first federal appellate decision addressing AI agents acting on users' behalf under the CFAA — and its commercial implications extend well beyond the two companies involved.
What the Court Actually Decided
Amazon sued Perplexity in late 2025 after sending a cease-and-desist letter demanding that Comet stop accessing Amazon's platform. A district court granted Amazon a preliminary injunction in March 2026. The Ninth Circuit stayed that order while it reviewed the case, then vacated it entirely on 4 August.
The court's reasoning was precise: under the CFAA, the relevant question is whether the party accessing the computer system had authorisation. Because users authorised Perplexity's agent to act on their behalf, the court found that Amazon was unlikely to prove the agent itself was an unauthorised accessor. The ruling does not resolve the full merits of Amazon's case, and it does not establish an unrestricted right for every AI agent to access every platform.
What it does establish is a significant precedent: a user's delegation to an AI agent may be legally equivalent to the user's own access, at least at the preliminary injunction stage under the CFAA.
The Deeper Question: Who Owns the Buying Journey?
Amazon's commercial model depends partly on controlling the interface through which products are ranked, promoted and purchased. Its advertising business — which generated over $56 billion in revenue in 2024 — relies on sponsored placements, merchandising algorithms and the discovery layer that sits between a customer's intent and a completed purchase.
Agentic commerce may unbundle that interface. When a customer delegates their buying intent to an external agent, the agent may:
- bypass sponsored product placements entirely;
- compare products based on structured data rather than visual merchandising;
- select based on price, availability and policy — not advertising investment;
- complete the transaction without the customer ever seeing Amazon's UI.
The platform becomes fulfilment and inventory infrastructure. The agent becomes the customer's interface. Retail advertising loses some influence over the final decision.
This is not a hypothetical future state. Perplexity's Comet is already doing this for users who choose to use it. The court has now said that doing so is not obviously illegal.
What This Means for Brands and Marketers
The ruling has three immediate implications for brands selling through Amazon and other platforms:
1. Structured product data becomes a competitive advantage. Agents evaluate products based on machine-readable signals: pricing accuracy, availability, specifications, reviews and policy clarity. Brands with incomplete or inconsistent product feeds are invisible to agents that rely on structured data rather than visual merchandising.
2. Advertising investment may not translate to agent visibility. Sponsored placements are a human-facing signal. An agent optimising for a user's stated criteria — lowest price for a specific specification, fastest delivery, best return policy — may ignore paid placements entirely. Brands need to win on the underlying product attributes, not only on advertising spend.
3. Platform access terms are under pressure. If courts continue to find that user-delegated agent access is equivalent to user access, platforms will need to revise their terms of service, API strategies and authentication flows to address agentic intermediaries explicitly. Brands that depend on platform exclusivity or proprietary data advantages may face new competitive dynamics.
The Agentic Commerce Readiness Framework
At Integrated.Social, we define Agentic Commerce Readiness across seven dimensions:
- Structured product data: complete, accurate, machine-readable feeds with real-time pricing and availability;
- Policy clarity: return, warranty, fulfilment and authentication policies stated in agent-parseable formats;
- Evidence quality: verified reviews, specifications and third-party validation that agents can evaluate;
- Transaction access: API and authentication flows that support delegated access without requiring human UI interaction;
- Agent-facing optimisation: content structured for AI evaluation, not only human persuasion;
- Attribution architecture: tracking that captures agent-intermediated journeys, not only direct sessions;
- Governance: clear policies on which agents may access which systems and under what conditions.
Brands that invest in these dimensions now are building the infrastructure for a commerce environment where agents, not websites, determine product consideration and purchase. Those that wait for the legal landscape to fully settle may find that the structural advantages have already accrued to competitors who moved earlier.
The Broader Pattern
The Perplexity ruling is one data point in a larger pattern. The Rezolve AI revenue evidence shows that agentic commerce is already generating real commercial outcomes. The OpenAI enterprise trust debate shows that agent governance is becoming a board-level concern. And the AEO and AI search visibility work we do with clients shows that the brands being cited and selected by AI systems are those that have invested in structured, evidence-led content — not those relying on legacy advertising infrastructure.
The court has not resolved who owns the buying journey. But it has made clear that the answer will not be determined by platform terms of service alone. It will be determined by which brands, agents and platforms build the most useful, trustworthy and accessible infrastructure for delegated commerce.
Modi Elnadi is the founder of Integrated.Social, a B2B AI marketing agency specialising in agentic AI lead generation, AEO/GEO and performance marketing. He has been working at the intersection of AI and commercial marketing since 2014.
