I mentioned it in a previous post 👉 https://exadvize.com/ai-in-e-commerce-when-it-chooses-for-us-who-bears-the-responsibility/
We explored a growing question: when AI makes choices for us… who carries the legal responsibility?
Since then, a first major clash has erupted, between Amazon and Perplexity, over its autonomous agent Comet.
🛒 Amazon sounds the alarm
➡️ Amazon sent a formal cease and desist letter to Perplexity AI.
Reason: its AI agent Comet may be performing autonomous purchases on Amazon using customer accounts behaving like a human user.
Amazon claims that Comet:
- acts as if it were a real shopper,
- without disclosing that it is automated,
- may not guarantee optimal choices (price, delivery, returns),
- generates additional investigation and security costs, and
- violates Amazon’s terms of use.
In short, Comet would be operating inside Amazon’s ecosystem without explicit authorization and without technical transparency.
🆚 Perplexity’s response: reframing the debate
Perplexity firmly rejects Amazon’s accusations, and in its public post, “Bullying is Not Innovation,” the company claims Amazon is:
- blocking the emergence of autonomous AI agents,
- locking down its ecosystem,
- limiting users’ freedom to choose how they browse and shop online.
They shift the question to a bigger issue:
👉 Do users have the right to rely on an AI agent to interact with a platform on their behalf?
⚖️ The legal issue
Two legal questions clearly emerge:
1️⃣ A potential violation of Amazon’s terms of use
If an AI agent performs actions on Amazon without meeting the rules set for human users:
- Who is bound by those terms?
- The user who provided account access?
- The developer of the AI agent? Or
- The agent itself (which the law does not recognize as a legal person)?
Because Comet imitates human behavior, a deeper question arises:
👉 Can a platform require all automated interactions to be explicitly declared as such?
2️⃣ A complete legal vacuum for autonomous AI agents
Comet does more than “assist.”, it acts, decides, and executes purchases on behalf of a user without human supervision at each step.
Yet current law provides no specific framework for:
- how AI agents may legally interact with online platforms,
- whether they must identify themselves,
- what behavior is unauthorized, or
- how responsibility is allocated.
👉 This is a true legal grey zone.
🧭 The rise of “agentic commerce”
This case signals that we are entering a new era where AI agents will:
➡️ browse,
➡️ compare,
➡️ and even purchase on behalf of users.
This raises several foundational questions:
- Can platforms refuse access to these agents?
- Must AI agents disclose their non-human nature?
- Who is liable for a flawed or unintended purchase?
- How do we ensure fairness and transparency in these algorithmic interactions?
Today, none of these questions have clear answers.
💬 A real legal test of the AI-agent era
The Amazon–Perplexity conflict isn’t just a commercial dispute.
It’s a wake-up call: current digital commerce rules were not designed to govern:
- autonomous AI decision-making,
- agent access to platforms, or
- responsibility for their actions.
👉 The real question is no longer who is right, but if we are ready for a legal framework where key commercial decisions are taken by autonomous AI agents ?
