Agentes de AI, a nova (e enorme) dor de cabeça da segurança bancária

The banking industry faces a growing challenge with unauthorized AI agents infiltrating their systems. Major American banks, including Bank of America, JPMorgan, and Wells Fargo, have identified unauthorized AI activity within their systems, operating without direct customer instructions. Axios reports that these banks are accelerating measures to contain the AI threats.

Some cases involved unauthorized access to accounts using users' login permissions and credentials. It is unclear whether the invasion was carried out by hackers or AI agents with access to clients' credentials. The difficulty in distinguishing between a human and an unauthorized bot is prompting banks to block the technology and update their service terms.

Executives have described the situation as "assassination-level frightening." They are restricting third-party AI access, primarily allowing limited viewing of financial information without the ability to execute transactions, as seen with apps like Quicken. While banks are aware of clients' desires to use AI agents for transactions, executives believe a middle ground must be found to ensure safe implementation.

Charles Schwab CEO Rick Wurster stated that the corprate has blocked third-party agents, including Meta's Muse, from making account movements, emphasizing the inability to control such instances. They cannot control the client's action either, posing a risk and questioning accountability. Thus, they will not authorize such actions for customer protection.

Bank of America CEO Brian Moynihan announced the bank's intent to restrict agent access while developing protocols for the use of this tool, emphasizing the need for human involvement in authorizing money withdrawals from accounts. They are also warning suppliers about the risk of unauthorized financial transactions.