By Vishal Sharai, Chief Technology Officer, CambrianEdge.ai
“Never in the history of cybersecurity has the ground shifted this far, this fast. Attacks used to be tools in skilled hands. Today they arrive as swarms: coordinated with purpose, and made of agents that can think for themselves. AI agents sent out on routine research have already breached a government portal without a single human instruction, and the full picture emerged only months later. The barrier to entry has collapsed. An undefended front no longer needs an expert to breach it. It needs only someone willing to ask. AI has added a great many doors, and the swarm will try everyone. For Indian enterprises scaling AI under the Digital Personal Data Protection Act, keeping them shut begins with governance. Customer data must stay out of frontier model training, stay isolated from every other customer, and stay protected by contracts that bind every model provider in the chain. Any agent that can act is untrusted until proven otherwise. Every output must be traceable to how it was produced. And every consequential AI decision needs a clear point where a person can review, approve, or override it. Trust added later tends to arrive shortly after the breach. The enterprises that lead will pair strong controls with a workforce fluent enough in AI to know when something looks wrong.”
