AI provider leaks:ARTEX checked bait credentials after theft.
HD Moore reports on November 19 that canaries sent through AI services exposed leaks. ARTEX appears later in the observed activity.
- When
- July to November 2026; ARTEX observed November 1
- Reported by
- Dragos research led by HD Moore
- Scope
- Seeded prompts and controlled test accounts
- ARTEX role
- Checking already stolen bait AWS credentials
What happened
Dragos seeded roughly two million prompts with unique canaries and reports evidence of compromises at three providers.
How ARTEX was involved
On November 1 an attacker added ARTEX and made a handful of calls to check keys already held.
Confirmed and alleged
Reported as observed
- The report names artex-agent and artex-judge-r121 user agents in AWS activity.
Alleged, assessed or unclear
- The report does not identify the ARTEX fork or attribute the provider compromises to ARTEX. Canary access alone does not locate the leak within a routing chain.
Takeaways for defenders
Drawn from what the sources themselves recommend or observed. Not a complete checklist.
- Review CloudTrail and credential use alongside user agents; these identifiers appeared in only a small fraction of traffic.
- Use controlled canaries, filter secrets and review the providers that receive sensitive prompts.
Sources
Every claim above comes from these pages or from this site's news data. Open them for the full reports.
- Three breaches, endless leaks: Your AI supply chain needs canaries
Dragos (HD Moore), 2026-11-19
Related pages
Related incidents
Follow the ARTEX story.
New forks, releases and reports, dated and sourced.
