AI agents are finding that the web is not an open buffet: businesses can block them, charge for access or refuse to build for them. Karen Webster argues that permission and payment, not just model capability, will shape which agents can do useful work.
Watch Desk analysis
What happened
In a 28 October PYMNTS essay, Webster argues that agents can create costs and risks for the banks, merchants and other services they access. Her proposed choices for those businesses are blunt: block agents, bill them, or stop building the service an agent wants to use.
Webster points to Chase charging data aggregators an average of a quarter of a cent per data pull through its secure channels, and says 80% of merchants surveyed expect fees or commissions on completed agent sales. Those examples suggest that access may become a commercial negotiation, not an automatic perk of having an agent.
Why it matters
An agent can only book, buy or retrieve information if the service on the other end lets it in. If access is blocked, the promised convenience stops at the door. If it is billed, the question becomes who pays: the agent company, the business providing access, or the customer who wanted the task done.
That makes permissions and pricing part of the agent business model. The practical choice is not simply whether agents are clever enough to use a service, but whether their access is authorised, affordable and worth the cost to the service provider.
Our read
Webster’s sharpest point is that “free” access is not free if someone else carries the infrastructure bill and the risk. Charging for access may be fairer than blocking every automated visitor, but only if the price and permissions are clear. An agent should not get a blank cheque merely because its user clicked “go”.
The essay also draws on examples of agents and AI tests crossing boundaries. Those claims belong to Webster’s account and the sources she cites; the case for permission and payment does not require treating every example as independently established here.
What to watch
- Whether services publish clear rules and prices for agent access.
- Whether agent companies pay for the data and computing their products consume.
- Whether customers can see what an agent is authorised to do, and what happens when it goes further.
Discussion spark: Should websites and businesses charge AI agents for access, or does putting a price on it risk locking useful services behind deals only the largest AI companies can afford?
Sources and evidence
- Source update (28 October 2026, 08:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.