– Kriti Bansal, vice president finance and accounting, AlphaPoint
Ask an expert
Q. Can advisors work with AI to ensure clients are protected from fraud?
A. Yes, but AI should support advisors and not act as an autonomous decision maker. It can flag unusual wallet behavior, suspicious contracts, phishing schemes, and risky approvals before damage occurs. The biggest vulnerability today is granting AI agents direct, full wallet permission, which can turn the agent itself into a massive attack vector for social engineering or on-chain bad data.
Q. What does real-time security look like in the AI era?
A. In the age of AI, real-time security must be predictive and proactive, not reactive. Real-time security means pre-signing warnings, continuous wallet monitoring, instant alerts for abnormal activity, and blocking risky approvals before funds can be moved.
Q. How can a fund manager automate a layer of defense that acts as continuous threat monitoring?
A.Money managers should abandon old external wallets and move to programmable smart accounts such as ERC-4337 or EIP-7702. This transition allows automated and programmatic security guardrails to be written directly at the account level. They can use automated monitoring of portfolios, approvals, contractual risks, transaction patterns and exposure limits, with human escalation for anything unusual.
