Anton Chuvakin on Faster Exploitation, Patching Bottlenecks and AI Security

Summary

Anton Chuvakin discusses shrinking exploitation windows, patching bottlenecks and AI security. He recommends mitigating unpatched vulnerabilities, checking logging coverage and applying least privilege to narrowly scoped AI agents.

Key points

• Chuvakin says exploitation timelines have accelerated but disputes the premise that exploitation within seconds is realistic.
• He identifies patch deployment as a major bottleneck, particularly for complex enterprise systems, and recommends layered mitigations where rapid patching is impractical.
• He recommends phishing-resistant MFA and comprehensive validation that basic controls are consistently deployed.
• Incident investigations can be hampered by missing DHCP, outbound or network-flow logs and insufficient retention. He suggests one year as a general retention rule of thumb, subject to circumstances.
• He cautions against making a CVE identifier mandatory for vulnerability workflows because vulnerabilities may be exploited before assignment or grouped under one identifier.
• For AI agents, he recommends narrow responsibilities and least-privilege access to tools and data, as well as identifying corporate and shadow agents.
• He advises patching edge devices and says security teams should maintain a structured plan for mitigating unpatched vulnerabilities.

People

Vendors