CogSec 2026 — Cognitive Security for Human–AI Systems
As generative and agentic AI become embedded in search, education, companionship, and decision support, a class of trust-and-safety risk is emerging that the field has no shared way to name, test, or govern. A single model output may be accurate and appropriately disclosed. The risk lies instead in how these systems shape attention, belief formation, memory, judgment, and agency across many interactions — and in how they reshape collective sensemaking.
CogSec proposes cognitive security — the integrity of the conditions under which people and groups reason for themselves — as a research agenda spanning HCOMP (AI-assisted decision-making, over- and under-reliance, oversight) and CI (misinformation, deliberation, civic decision-making).
Call for participation
We welcome short papers, position papers, demos, and datasets:
2–6 pages, excluding references, in the ACM Primary Article Template — or an extended abstract plus a demo description
Non-archival, so work-in-progress is welcome and you keep the rights to publish the full version elsewhere
Single-blind review — no need to anonymize
Accepted work appears as a lightning talk and/or poster
Topics: over- and under-reliance on AI · cognitive vulnerabilities · persuasion and manipulation · misinformation and prebunking · provenance and deepfake detection · frictional design · governance — and adjacent work of all kinds.
Deadline: August 17, 2026 (AoE) · Notifications: August 20 · Workshop: September 27
👉 Submit via EasyChair: https://easychair.org/conferences/?conf=cogsec2026
📄 Full call for participation: https://noetx.github.io/cogsec-2026/
Attending
The workshop is hybrid — join in person in Washington, DC or remotely.
⚠️ Please note: RSVP'ing here is not conference registration. Attending CogSec means registering for HCOMP / CI 2026 at the workshop-day rate. We'll post the registration link here as soon as the conference publishes it.
Questions? Email vastu@stanford.edu.