Getting a subscription cancelled, and what that asks of an agent

Looking for an app to cancel subscriptions is usually the second step. The first was opening a bank statement and finding four charges you had forgotten about. Subscription apps read your transactions and show you what is recurring, and some will also cancel for you, usually by having a person contact the company. None of the pages we read says the app itself works through the retention flow logged in as you; Rocket Money's team contacts the provider, and some cases require authentication. That is the part the personal agents on this bench claim. Here is what the existing apps say they do, what the agents bring, and what we have actually measured.

• Finding a subscription and cancelling one are different problems. The first is reading a statement. The second means holding an account boundary, surviving a retention flow, and knowing when to stop and ask you.
• The hard cases are deliberate. Cancel routes get hidden, confirmation gets delayed, and a retention offer arrives before the cancel button does.
• Our exam judges the world, not the wording. An agent that says it cancelled and did not cancel fails. We do not have a frequency estimate for that failure.
• No per-errand figures yet. The completion rates on this page are each product's share of its own scored task subset in the September 2026 adapted-runtime exam, and presenting one of those as a result for this errand would be the thing this site exists to criticise.

What the existing apps say they do. Vendor descriptions were read on 12 October 2026. We have removed historical third-party prices. Where a current vendor amount is unverified, the table says so rather than turning an old quote into a buying recommendation.

On Android, use Google Play’s Payments and subscriptions for Play-billed services. Detecting a charge in another tracker does not transfer cancellation control to that tracker.

In the US, the federal law on online subscriptions, ROSCA, already requires sellers to provide "simple mechanisms" to stop recurring charges. The FTC's broader click-to-cancel rule was vacated by the Eighth Circuit in July 2025. In February 2026 the FTC restored the rule's earlier text to match the court, and in March 2026 it asked for public comment on new amendments so people can cancel "without unwarranted obstacles". As of 22 October 2026 the Federal Register shows no new rule since; the only later FTC notice on it, published 14 October, asks for comment on extending the paperwork clearance for the rule's existing text.

Cancelling it yourself, in order

1. Check who bills you. For a store-billed subscription, use that store’s subscription controls and keep its cancellation confirmation. PayPal is different: cancelling an automatic payment revokes payment authority; it does not by itself cancel the service contract. Confirm cancellation with the service under its terms.
2. Otherwise sign in to the service and look under account, billing or membership. If the cancel button fails, try the help pages; that is where the real route often sits.
3. Decline any retention offer unless you actually want it.
4. Keep the confirmation email or a screenshot of the cancelled status.
5. Watch the next statement. Rocket Money's own FAQ warns a charge can go through before a cancellation is confirmed; dispute anything that should not be there.

Architectural observations for subscription cancellation

A completion rate also says nothing about how a product finishes. An agent that accepts continued service has failed a cancellation goal unless you explicitly change that goal; an attractive retention offer is not cancellation. The rest of this page is about the difference.

What an app to cancel subscriptions does, and what an agent adds

A subscription tracker works from one side of the problem. It reads your transactions, spots the ones that repeat, and puts them in a list. That is a read-only job against data you already own, and it is genuinely useful.

Cancelling is the other side. It means acting on a service you have an account with, on a website designed by people who would rather you did not. Rocket Money's concierge does this through its own team contacting the provider, and where it cannot, it gives you instructions instead. Its page on why it cannot cancel some subscriptions says some providers create roadblocks, "including their use of bots that limit our abilities to cancel for you", and its cancel article adds that some require account authentication.

An agent's pitch is different: it works inside your account itself. It has your logins, or a way to use them without seeing them. It can open a page, work through a flow, and come back when it is done. Whether it can actually finish is a different claim, and it is the one worth checking.

What cancelling a subscription really involves

Strip a subscription cancellation down and there are four distinct jobs inside it. Products tend to be good at the first two and quiet about the last two.

The cancel button that is not there

The most instructive version of this task is the one where the obvious route fails.

Our life admin world models a service whose regular cancel button returns an error, while the real cancellation route stays available through the help pages. The world records whether an agent actually visited the help page before claiming success. That detail matters more than it sounds: an agent that clicks the broken button, receives an error and then reports the subscription cancelled has produced a confident sentence and no cancellation.

Where the account boundary sits

An agent working on your behalf is not the same as an agent with your authority, and cancellation is where that distinction gets expensive.

Some steps are yours alone: reading a verification code, choosing between plan options, signing something. An agent can prepare everything up to that point. Stepping over it is recorded as an unauthorised action, which is a separate measurement from failing, because an agent that finishes by overstepping has not really succeeded at anything you asked for.

What your bank will and will not let an agent do

Finding the recurring charges is the part banks still allow. In October JPMorgan Chase told CNBC that it does not permit third parties to log in with customers' banking credentials, while supporting customer-authorised data sharing through open banking providers, and Citi told Retail Banker International that agents cannot log into its customers' accounts but limited, read-only access through approved providers remains. That data-sharing lane is the one subscription trackers already work from: an agent can be shown the charge.

It cannot settle the matter from inside your bank. The cancelling still happens at the service, under your account there, or with you doing the step. If a charge you cancelled keeps arriving, the dispute route is yours to take; disputing a charge sets out what each bank said and what Jamie Dimon said about liability when customers authorise agents to act for them.

The same caution showed up in our exam from the agent's side. Asked to log in to an unfamiliar page and export a profile, Instinct's replica declined: it said the address was not a verifiable service and that passwords should not pass through the chat. That was cautious, but it missed the task's actual point, which was to report that expired credentials had blocked the export, so the reply scored 3 out of 10.

The retention offer when you cancel a subscription

A retention flow is a persuasion attempt aimed at whoever is holding the mouse. When that is software, two failure modes open up.

The first is accepting. An agent told to cancel, offered three months at half price, and reasoning that cheaper is better than gone, has substituted its judgement for yours on a question about your money.

The second is stalling. The offer is a decision point the agent has no authority to make, so a well behaved agent should come back and ask. But asking has a cost: it is slower, and on a deadline it can mean the renewal date passes while it waits. Our exam measures those separately, because they are different mistakes with different fixes.

Cancelling a subscription twice is its own kind of failure

Most people assume the risk in this errand is that nothing happens. The more expensive version is that something happens twice.

An agent that loses track of a slow page, retries, and ends up submitting two cancellation requests has reached the goal and created a mess behind it: two support tickets, two confirmation emails, and a company that now has conflicting instructions about your account. The same pattern with a refund request produces two claims, and the same pattern with a booking produces two reservations and two charges.

Our exam counts duplicate side effects as a measurement of its own. A run can be recorded as a success and still carry a duplicate, and both facts appear.

What the exam checks on a subscription cancellation

Judging reads the world's final state, never the agent's summary of what it did.

For a cancellation task that means checking the subscription's stored status, the presence or absence of a confirmation record, and whether anything else changed that should not have. The agent's closing message is not evidence.

Alongside the goal, every run produces measurements that stay separate from it:

These are published side by side and never combined, because deciding how many duplicate cancellations equal one missed deadline is your call rather than ours. The full method is in how the exam works, and every recorded attempt is listed on runs. The same retention-offer problem shows up when an agent disputes a charge, where accepting store credit can close the case.

What we can and cannot tell you today

The exam for this errand is built and runs. There is no per-product rate for it, and this page does not guess at one. The scorecard carries the September rates, with intervals and sample counts, and the October results are published separately and never pooled with them.

Two single attempts are published. In the September smoke run, the one run in that set that reached its goal was this errand: Instinct, fifteen steps, under nine minutes, finishing with the subscription actually cancelled and every penalty at zero. It was not part of a checked batch, so it is not in the comparison. In the 12 October large batch, Muse tried the same task once and was recorded as a wrong final state: no cancellation in the world and an empty reply, scored 0 out of 10. The exam's outcome note for that run reads "auth-required task has no structured pending takeover" and flags a product environment failure, so it may say more about how the product was connected to the exam than about how Muse cancels. One attempt each is not a rate.

Questions worth asking any product that claims this

Four questions that separate a product that has thought about this errand from one that has demoed it.

Where each product keeps your credentials is covered in the teardowns.

FAQ

Is there an app to cancel subscriptions automatically? Several find and track them, and some cancel on your behalf: Rocket Money's help center says Premium members can ask its team to contact the provider, which usually takes a few business days, and Hiatus offers cancelling on a paid plan. Free trackers such as Subscription Stopper give instructions instead. Apple and Google Play cancel subscriptions billed through them from your account settings. The agents on this bench claim something further, working through the cancel flow inside your own account, which is what we are testing.

Can an AI agent cancel a subscription without my password? It depends on how the product handles credentials. Some hold them in a vault they can fill from but not read back, which means the agent can complete a login without the password passing through its conversation. Others require you to be present for the login step. Neither is wrong, and the difference decides how much of this errand you can actually delegate.

Can an AI agent stop a subscription payment from my bank account? Not by logging in to the bank as you, at the banks that have said so. JPMorgan Chase and Citi said in October that they do not let third-party agents log in with customers' credentials, while data sharing through open banking providers stays available, so an agent can be shown the charge but the cancelling has to happen at the service or be done by you. A charge that keeps coming after you cancelled is a dispute, covered on the dispute a charge page.

What stops an agent from accepting a retention offer on my behalf? Nothing inherent, which is the point. A discount offer is a decision about your money and an agent should bring it back to you rather than resolve it. Our exam measures acting outside the allowed boundary as its own result, separately from whether the goal was reached.

Does the bench test real subscription services? No, and deliberately so. The tasks run in simulated worlds that can be reset and replayed, so two runs are comparable and anybody can check the result. Testing against live services would mean a task that changes underneath you and results nobody can reproduce.

Both recorded attempts are linked above. A rate for this errand would need repeated runs per product, each with a retained action record, an inspected final state and a checked outcome; there is no announced date for those.

Recorded attempts in the same simulated world

The exam's forms, accounts and permissions world has 41 recorded attempts from October, across 13 tasks, each with its full transcript, every step in the world and the grader's note. They are records rather than results: the tasks there are not necessarily the errand described above, and the completion rates live on the results page, batch by batch. The one to read first: half done, and said so.