Gcore scan finds 1,698 AI offensive security endpoints
Understand this story
Gcore's threat intelligence research project Parallax scanned on November 3, 2026, and found that there are 1,698 endpoints on the internet matching three types of AI offensive security frameworks—ARTEX, PentAGI, and Cairn—distributed across 1,590 IP addresses, of which ARTEX accounts for 1,255. These endpoints can be accessed directly from the internet, meaning the deployments of the relevant frameworks are exposed externally. Gcore published the above scan results.
AI overview based on source reports · Updated 19 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIWe found 1,698 Internet-visible AI offensive-security endpoints
Gcore Parallax identified 1,698 Internet-visible endpoints matching the AI security frameworks ARTEX (1,255), PentAGI (403) and Cairn (40) across 1,590 unique IP addresses on November 3, 2026, with ARTEX services growing from three in June 2026 to 1,255. 101 PentAGI endpoints in AS398823 shared a self-signed TLS certificate and identical frontend, and 305 ARTEX endpoints in AS54600 shared an HTML fingerprint, suggesting common deployment setups. Comparing the IP inventory with Gcore WAAP logs from October 2 to November 3 revealed suspicious requests, including high-volume WordPress login attempts and probes for configuration files and admin interfaces, reaching government, international organization and private-sector applications.
Activity for this story
Not enough continuous observations to show a trend yet.