SEC536: Adversarial AI - Penetration Testing AI Systems


Experience SANS training through course previews.
Learn MoreLet us help.
Contact usBecome a member for instant access to our free resources.
Sign UpWe're here to help.
Contact UsKey findings:
The survey's central argument is that identity security has entered a new phase where the challenge is no longer buying tools but operationalizing them. Organizations successfully hardened human login security with MFA and SSO, but the attack surface shifted to what happens after authentication succeeds: ungoverned sessions, static service account credentials, and autonomous AI agents entering production faster than governance frameworks can be built. Closing the gap between detection and containment, not further tool adoption, is what the data points to as the defining challenge for 2026. Respondents were primarily based in the United States (85%), with additional representation from Europe, Canada, Africa, Asia, and Australia. The top industries represented were cybersecurity, technology, banking and finance, and government, with the largest organizational segment falling in the 100–1,000 employee range. Respondents spanned the organizational hierarchy, from security engineers and SOC analysts to CISOs and security VPs.
Overview Identity has become the new battleground. From SaaS to cloud to legacy Active Directory, it is now the central control point—and attackers know it.

The survey identifies a "deployment vs. resilience" gap: 85% of organizations deploy ITDR tools, but 55% still experienced an identity-related compromise in the past 12 months, because most organizations can detect attacks faster than they can contain them.
Poorly. 76% report growth in non-human identities, but 92% fail to rotate the majority of their NHI credentials on a 90-day cycle, and only 8% rotate more than 75% of them that often.
No. Credential phishing accounts for only 35% of attacks; most incidents now involve techniques that bypass the login entirely, such as compromised browsers (27%) and session token hijacking (23%).
Not adequately. 73% of organizations are deploying agentic AI or automations that require credentials, but no single governance control, such as approvals or audit trails, is used by more than 40% of organizations.
Reputation and brand damage, cited by 44% of organizations as the primary outcome, ahead of service disruption (40%) and loss of shareholder value (30%).


Rich Greene, SANS Senior Solutions Engineer and SEC301 author, brings 20+ years of cyber operations and teaching experience to the classroom. With 15+ GIAC certifications and a passion for mentorship, he equips defenders with real-world confidence and skill.
Read more about Rich Greene















