As noted by CRN, Netskope co-founder and CEO Sanjay Beri emphasized that businesses cannot rely on regulators or AI platform providers to manage the risks associated with AI agents potentially behaving erratically. During a special episode of the Security or Else! podcast, Beri highlighted the urgent need for companies to implement their own safeguards against rogue AI.
Beri stated that companies must proactively secure their AI agents, as no external entity will legislate against AI going rogue or producing harmful content. He stressed that businesses, in collaboration with vendors and partners, need to take responsibility for preventing unexpected AI behaviors that could lead to data breaches or system compromises. This involves identifying unsanctioned tools, implementing real-time controls, and establishing governance. Beri suggested that an AI "kill switch" could involve revoking an agent's identity or halting its activity. He advocated for a zero-trust approach to AI models, urging companies not to assume inherent safety from platform providers. For solution and service providers, this presents a significant opportunity to act as trusted advisors, offering expertise and guidance to customers navigating the complexities of AI security.
Source: CRN
