RESEARCH RECORD

Evaluating AI cyber capabilities with crowdsourced elicitation

Artem Petrov · Dmitrii Volkov · Palisade Research

Authors report competitive performance from crowdsourced agent configurations.Human comparison enables task-duration analysis.

Research question

Can crowdsourced elicitation complement internal cyber evaluations?

Open AI tracks at two CTF events collect elicitation attempts and human comparison data.

Authors report competitive performance from crowdsourced agent configurations.

Human comparison enables task-duration analysis.

Institutional summary only; competition rank is not autonomous real-world intrusion probability. Full methods remain pending.

No structured numerical extraction is available in this reading package.

Reading scope & review

Institutional abstract/summary only; full linked paper methods/results not reviewed

Model-assisted reading; independent evidence review pending.