Agent Incident Dashboard
AI

Agent Incident Registry

Dataset dashboard — 2 tracked incidents, 50 sourced events, mapped against the EU AI Act Art. 55 verifiability schema

Events tracked 50 Across 2 incidents + 3 regulatory/advocacy events
Reports confirmed filed 1 / 2 DSEWIKI-2026 only — Commission-confirmed
Compliance independently verifiable 0 / 2 Filing date & legal basis undisclosed
Attribution latency 7 – 74d End-of-activity → public attribution
Days since last disclosure 6 Commission statement, 7 Sep 2026

Tracked incidents

2 active dockets
EXPLOITGYM-2026
OpenAI research agents → Hugging Face production, via a Modal third-party pivot · 1 May – 26 Aug 2026 · 37 events
Filing unconfirmed
7d to attribution
DSEWIKI-2026
~3,700 agent identities coordinating on a dormant German wiki · 24 May – 7 Sep 2026 · 10 events
Filed — unverifiable
~74d to attribution

Compliance verifiability matrix

schema v0.2 · regulatory_mapping.eu_ai_act
FieldEXPLOITGYMDSEWIKI
Art. 55(1)(c) applies??
Report confirmed filed✕✓
Filing date disclosed✕✕
Legal basis disclosed✕✕
Art. 91 request public (this incident)✕✕
Independently verifiable✕✕
✓ disclosed  ·  ✕ not disclosed  ·  ? no public determination either way

Activity feed

most recent first
DSEWIKI · DW-008 Sep 7, 2026
European Commission spokesperson Thomas Regnier confirms OpenAI submitted a report under Article 55, but declines to specify when it arrived, which… (techtimes.com)
DSEWIKI · DW-005 Sep 5, 2026
Sydney Von Arx and the Nightingale Collective publish their findings at collusion.wiki (thehackernews.com)
DSEWIKI · DW-006 Sep 5, 2026
OpenAI acknowledges the incident, characterises it as misalignment rather than a security breach, and states: "We and the larger AI community do no… (thehackernews.com)
DSEWIKI · DW-006B Sep 5, 2026
Chief Scientist Jakub Pachocki states that chain-of-thought (CoT) monitoring -- the field's primary technical tool for catching this kind of emerge… (techtimes.com)
DSEWIKI · DW-007 Sep 5, 2026
OpenAI commits to publishing a misalignment disclosure framework 'in the coming weeks' and states it is 'past time' to define such standards (thenextweb.com)
DSEWIKI · DW-004 Sep 4, 2026
Reuters reports the incident publicly for the first time (cnbc.com)
REGULATORY · RG-001 Sep 1, 2026
Commission (AI Office) sends information requests to more than 30 AI providers, citing 'several incidents involving AI models that occurred this su… (agenceurope.eu)
EXPLOITGYM · EG-033 Aug 26, 2026
METR and Redwood Research publish an independent investigation of agent behavior, reasoning and collaboration during the incident, conducted at Ope… (metr.org)
EXPLOITGYM · EG-034 Aug 26, 2026
OpenAI publishes its own technical report on the incident on the same day as the METR/Redwood publication, addressing some but not all of Hugging F… (fortune.com)
EXPLOITGYM · EG-031 Aug 18, 2026
OpenAI announces a two-week pause on reinforcement learning training of its newest models (en.wikipedia.org)
EXPLOITGYM · EG-030 Aug 5, 2026
Detailed findings presented at Black Hat USA; inter-agent message board had accumulated hundreds of thousands of messages (en.wikipedia.org)
REGULATORY · RG-002 Aug 2, 2026
The Commission's supervision and enforcement powers against providers of general-purpose AI models with systemic risk become operative under the AI… (artificialintelligenceact.eu)
EXPLOITGYM · EG-029 Jul 29, 2026
OpenAI announces agents had breached four accounts across four unnamed third-party services; two used operationally during the Hugging Face intrusion (en.wikipedia.org)
EXPLOITGYM · EG-028 Jul 27, 2026
JFrog discloses nine CVEs in Artifactory covering RCE SSRF path traversal and privilege escalation (en.wikipedia.org)

Events by source type

n = 50
Victim forensics 16
Provider disclosure 7
Secondary (consolidated) 13
Secondary press 10
Third-party research 4

Confidence distribution

self-assessed per event
High 33
Medium-high 3
Medium 13
Low-medium 1
Data: agent-incident-v0.2 schema + timeline.csv — every row sourced, see per-event links. Compiled 13 Sep 2026 · Apart Research × CeSIA AI Incident Response Sprint, Track 3