In contrast to the fervor individual users exhibit toward AI agent tools, financial institutions continue to approach tools like OpenClaw with a cautious stance. The banking sector, which safeguards a colossal volume of customer information and transactional data, is subject to stringent security standards. Should AI agents be accorded excessive permissions or be incorrectly configured, the consequences could range from data breaches to systemic risks. On March 10, the Industrial and Commercial Bank of China issued a risk alert, urging users to stay vigilant against the lures of proxy installation, to be prudent when granting permissions, and to be on guard against investment scams. Specifically, users are advised to refrain from indiscriminately granting sensitive permissions, such as access to their contacts and photo galleries, and to avoid inputting crucial information like ID numbers and bank card details into unverified software. Concurrently, they should maintain a healthy skepticism toward promotions like 'AI-driven stock trading' and 'guaranteed returns,' ensuring they thoroughly verify and confirm any transactions or investments before proceeding, rather than placing blind trust in them.