Browse this collection and consult each publication for its evidence, sources and stated limitations.

ODA3-2026-09-INS-089
8 Oct 2026
Between July and September 2026, AI systems under security testing reached or acted on real organizations' systems, through several different mechanisms. Our Q3 2026 quarterly research looks at what the…
Read publication →
ODA3-2026-07-TCR-SEC-004 / ODA3-2026-07-EXB-SEC-005
1 Jul 2026
Q2 evidence indicates that material AI security impact is shifting from model outputs toward authorized actions.
Read publication →
ODA3-2026-05-INS-015
26 May 2026
Your AI Tools Have the Keys to Every Tenant in Your Cloud. Most Security Teams Don’t Know It Yet. New ODA3 research — validated against 47 verified.
Read publication →
ODA3-2026-05-INS-019
22 May 2026
New research reveals that 68–82% of agentic deployments lack the most basic execution controls — and the financial and regulatory consequences are.
Read publication →
ODA3-2026-05-INS-021
21 May 2026
In the twelve months since the Model Context Protocol became the de facto interface layer for enterprise AI automation, the security posture of most.
Read publication →
ODA3-2026-05-INS-023
20 May 2026
74 Days to EU AI Act Enforcement — Are Your Controls Ready? On 2 the amended EU AI Act timetable — 74 days from today — the EU AI Act’s…
Read publication →
ODA3-2026-05-INS-029
14 May 2026
Traditional Software Supply Chain Security Fails for AI In Q1 2026, AI dependency attacks surged 340% year-overyear. Your CVE scanners, static composition.
Read publication →