In the wild
First threat actor observed with a zero-day exploit assessed to have been developed with AI
AIR-2026-0108 AIR record · 2026-05-gtig-ai-developed-zero-day
From the supporting source
For the first time, GTIG has identified a threat actor using a zero-day exploit that we believe was developed with AI. The criminal threat actor planned to use it in a mass exploitation operation but our proactive counter discovery may have prevented its use.cloud.google.com
The record
Disclosed
2026-05-12
Vendor
undisclosed
Product
A popular open-source web-based system administration tool (2FA bypass)
Risk domain
Access control and permissions › Privilege escalation
System type
Generative model
Who acted
Human acted on output — AI helped develop the exploit; the actor planned to deploy it
Attack surface
Other
Autonomy level
Semi-autonomous
Initial vector
Ai assisted vulnerability research
Impact type
Authentication bypass
Confidence
High
Submitted by
EAI Internal
Crosswalk
- OWASP Agentic Top 10 (2026)
- ASI03 – Identity and Privilege Abuse
- MITRE ATLAS
- AML.T0055 – Unsecured Credentials
- NIST AI RMF
- GOVERN 6.1
- EU AI Act
- Article 13
- ISO/IEC
- TR 24028; 42001; 23894
- AIUC-1
- B. Security
- Scope
- tool misuse
- Mitigation
- Apply least-privilege principle; Regular policy audits.
- Taxonomy
- Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
Planned mass-exploitation operation; Google assesses its counter-discovery may have prevented use
Capability abused
Defense & remediation
Recorded defense
Google proactive counter-discovery
Guardrail kinds
Unknown
Guardrail outcome
Silent
Fix shipped
Not stated
Taxonomy & references
Curator notes
The single most-cited 2026 data point for 'AI-driven vulnerability discovery used offensively'. Note GTIG hedges twice - 'we believe was developed with AI' and 'may have prevented its use' - and does not name the product or publish the exploit, so impact_realized is false. No independent corroboration exists.
Supporting sources
Cite as
Enkrypt AI. "First threat actor observed with a zero-day exploit assessed to have been developed with AI". AIR, AIR-2026-0108. https://agentincident.vercel.app/i/AIR-2026-0108/ (accessed 2026-10-19).