Could a company tell when an AI agent cancels a subscription instead of a person — and charge each one differently?

This explores whether a seller could spot an AI agent cancelling a subscription and respond with retention offers, different prices or friction, so the old pricing model survives for everyone else. The corpus doesn't test this directly, so treat what follows as an inference from nearby work rather than a finding. The stakes come from Will AI agents destroy the negative option billing business model?. Subscription pricing quietly depends on an 'inertia subsidy': people who forget to cancel pay for the deals that vigilant customers get. An agent that never forgets breaks that arrangement. So spotting agents is really about whether sellers can keep charging the forgetful while handling the vigilant separately.

There are reasons to think agents would be easy to spot. Shopping agents don't react like people. Under the same price and rating changes, they swing much harder and fall into rule-like patterns, where humans shift only a little Do shopping agents respond more strongly to price and rating cues?. Agents also aren't short on time. They can run many negotiations at once and keep them going indefinitely Why might AI agents bargain more patiently than humans do?. A cancellation that arrives on schedule, across dozens of services, with the same script each time, looks like a machine. Sellers are also good at reading buyers. A seller-side model recovered a buyer's willingness to pay almost perfectly from the conversation alone Can a seller infer what a buyer really wants to pay?.

There are also reasons spotting them could be hard. LLM agents copy human-like decision biases, learned from their training data rather than from human memory or attention limits Where do LLM choice biases actually come from?. In principle, an agent could be tuned to look human. That leads to the deeper problem: spotting agents starts an arms race. Will sellers game Google's new agentic commerce protocol? argues that sellers have always hidden their real prices from machine readers, from semantic-web crawlers to search engines. Treating agents differently at cancellation time is the same move. Buyers would respond by making their agents harder to detect.

The less obvious point is this. Even if sellers could spot agents perfectly, it might not save their pricing. The subsidy came from customers who weren't paying attention, and an agent is attention you can hand off. Once enough customers use one, there are too few forgetful payers left to fund the deals, however well sellers sort the cancellations. The more likely outcome is pricing that shifts in other ways, not detection. One example is the agent-to-agent price drift in Do LLM pricing agents autonomously collude in repeated auctions?, where pricing agents settled on higher-than-competitive prices without being told to.