Cybersecurity Experts Warn Against Autonomous AI and Legacy Patching

Cybersecurity executives are calling for a fundamental shift in defense strategies as artificial intelligence accelerates the pace of digital threats. Jamil Farshchi, Chief Technology Officer of Equifax, warned that the release of Anthropic's Claude Mythos model allows attackers to identify and exploit vulnerabilities in hours, rendering legacy IT patching processes that take months obsolete. Farshchi argues that traditional static scoring systems are no longer effective and advocates for prioritized risk models that focus on actual attack paths and business impact.

While AI provides new tools for automation, other experts caution against removing human oversight. Moe Ibrahim, Vice President of Sales Engineering for Asia Pacific and Japan at Exabeam, argues that pursuing a fully autonomous security operations center is a flawed strategy. Ibrahim asserts that human decision-making is essential to ensure security actions align with business priorities and regulatory requirements without disrupting critical processes.

Together, these perspectives suggest an emerging model for the agentic enterprise. This approach emphasizes a human-on-the-loop system where AI handles rapid execution and analysis, but humans maintain governance, explainability, and final policy authority to manage systemic risks.