Researchers at Spectrum Security say they found 2,335 internet-exposed consoles linked to offensive AI and agentic security frameworks, including ARTEX, PentAGI and CyberStrikeAI. Their investigation also identified three previously undocumented agent tools, making this more than a fresh count of familiar software.
Watch Desk analysis
What happened
The researchers say they traced exposed installations and variants, including stock setups and systems in distributed hosting fleets. They used shared TLS certificates and private build strings to map deployments and track distinct operator fleets, according to their report, “Swarm Chasing”.
The report describes frameworks such as ARTEX and PentAGI as being used across campaigns. The supplied account does not establish what each exposed console was doing, or that every exposed instance was being used in an attack.
Why it matters
Exposed management consoles can give researchers a view of how these tools are deployed and adapted. The discovery of renamed and forked variants, alongside three undocumented tools, also suggests defenders may be looking for recognisable names while deployments change around them.
That is a useful warning, not proof that every system identified was compromised or involved in malicious activity. Spectrum Security’s findings are the evidence for the count and the reported tool use; the distinction between exposure and active abuse matters.
Our read
The notable figure is 2,335, but the more consequential finding may be the ecosystem around it: forks, renamed frameworks and tools researchers had not documented before. Security teams should treat the report as a reason to check for exposed consoles and unexpected variants, not as a tally of confirmed intrusions. The internet remains generous with dashboards nobody meant to leave on the front porch.
What to watch
- Whether Spectrum Security publishes further technical details on the three undocumented tools.
- Whether defenders report finding these frameworks in their own environments.
- How quickly operators change exposed installations or move to renamed and forked versions.
Discussion spark: Should organisations treat an exposed offensive AI console as an urgent security incident, or wait for evidence that it has been used?
Sources and evidence
- Swarm Chasing: Hunting ARTEX and Other Agentic Offensive Frameworks in the Wild (18 November 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.