Financial institutions remain cautious overall towards AI agent tools like OpenClaw

Odaily News Compared to the enthusiastic experimentation by individual users, financial institutions overall remain cautious towards AI agent tools such as OpenClaw. Multiple interviewees pointed out that banking systems handle vast amounts of customer information and transaction data, requiring extremely high security standards. If AI agents are granted excessive permissions or configured improperly, they could lead to data breaches or system risks. "The volume and sensitivity of bank data far exceed that of typical user environments; once a vulnerability appears, the impact scope would be much larger," said a technology professional from a joint-stock bank. On March 10th, ICBC issued a risk notice related to OpenClaw, warning users to be wary of installation traps, exercise caution regarding authorization risks, and guard against investment scams. On one hand, do not casually grant sensitive permissions such as access to contacts, photo albums, or folders, and especially do not enter information like ID numbers, bank card numbers, or payment passwords into unfamiliar software. On the other hand, remain vigilant against marketing claims like "AI stock trading," "guaranteed profits," or "high returns with low risk." For any matter involving transfers, top-ups, or investments, always verify thoroughly and do not trust easily. (Yicai)