AI Security Risks Grow as Agentic Attacks Accelerate

AI security is quickly becoming a much bigger problem for channel partners. It's also becoming a much bigger services opportunity for partners. Two headlines this week show why that is. First, Google is warning that attackers have moved far beyond simply using generative AI to develop malicious code or to write better than average fishing emails. In one attack, Google tracked an autonomous suite of agents coordinated planned and executed a credential harvesting campaign in under 6 hours and was ultimately successful in stealing thousands of credentials.

Recent warnings from the UK's National Cyber Security Center point to shadow AI as an ongoing and increasing risk. In research that it cited from Microsoft, 71% of employees across the UK admitted to using AI tooling that their leadership hadn't approved or didn't know about. Now, taken together, those headlines show that AI security is becoming an issue on both sides of the firewall. Your attackers and your employees are posing risks to you and your customers.

So, what should you as partners do about it? Start building AI governance into your existing security practice right now. Think about what it would mean for customers not just to introduce new tooling but to get a handle on the identities, threats, and data policies that they already have in place whether they know it or not. Looking ahead, I'd expect AI governance and AI security to become an interwoven conversation. Customers aren't just going to ask partners, "How do I use AI?" Increasingly, they're going to ask, "How do I know what AI is doing within my business and how can I get better at controlling it?" And that is an opportunity for you to continue to act as a trusted advisor.

Head to channelinsider.com to subscribe to our weekly newsletter and get the how-tos and news you need to move your business forward delivered right to you and come back here next week for another inside the headlines. I'm Victoria Durgan.

This transcript was generated automatically from the video's captions and may contain errors.