India's CERT-In issues AI safety guidelines restricting automatic permission assignment
Technology
India's cybersecurity agency, CERT-In, just dropped fresh guidelines for using AI safely, especially when the stakes are high.
The main idea? Don't let AI bots run wild with user permissions.
By stopping AIs from automatically picking up your access rights, they're making it much harder for hackers to sneak in malicious commands and hijack systems.
Major AI decisions require human approval
The new rules also say every big decision made by an AI needs a human OK before it goes through.
Auditors have to watch out for weak spots, keep track of what the AI does, and make sure there's a way to hit pause if things go sideways.
With more businesses jumping on the AI bandwagon, these steps are all about keeping data safe and making sure humans stay in control.