The reported measure
Major US banks have begun introducing restrictions on AI agents' access to customer accounts. That is what Agenda Digitale reports, placing the development within the framework of digital markets and financial services. The source does not indicate which institutions have adopted the measure or the technical or contractual ways in which the limit is applied.
The available information therefore describes a trend, not a comprehensive measure: the material does not show a number of banks involved, a common start date, or a supervisory authority that imposed the choice. The scope of the intervention remains to be verified for now.
Why the issue matters
The fundamental point concerns who is liable when an AI agent carries out unwanted transactions. The source explicitly cites the case of unauthorized bank transfers as an open question: establishing liability among the customer, the bank and the technology provider is not straightforward. Agenda Digitale does not indicate which legal regime applies or report positions from authorities or courts.
A second dimension is added to this, which the source defines as systemic risk: automated movements of deposits, if carried out by software agents, could amplify the effects of simultaneous behavior. This is a risk raised by the article, not an event that has occurred or a quantified estimate.
What remains to be established
The material does not clarify whether the limitations are technical measures, such as blocking external agents' access to accounts, or contractual clauses. It does not indicate whether the banks have publicly communicated the criteria adopted, nor whether there are exceptions for certain categories of customers or operators.
Elements on practical consequences are also missing: it is not known whether the restrictions concern only read access, the sending of payment instructions, or both. On these aspects the source is silent, and any clarification would require direct confirmation from the institutions.
The information picture
The news item, signed by Agenda Digitale, should be read as a signal of a trend: the available article text is largely taken up by website navigation elements and contains no additional data beyond the excerpt. Questions about liability and systemic risks are raised, but they do not receive a documented answer in the material.
For this reason the matter remains one to follow: confirmations are needed on the institutions involved, on how the limits are applied and on any decisions by regulatory authorities. Until then, the only fact reported is that some major US banks have begun limiting AI agents' access to accounts.
