AdvML-Frontiers-CoTMA 2026

Deadlines
NLP/CORE Unranked

AdvML-Frontiers-CoTMA 2026

AdvML-Frontiers × CoTMA: From Model Security to Compositional Threats in Multi-Agent AI Systems

Oct 09 2026San Francisco, California, United StatesOfficial conference site Site reachable

AdvML-Frontiers × CoTMA @ COLM 2026 is a joint workshop uniting two communities—AdvML-Frontiers and CoTMA (Compositional Threats in Multi-Agent AI Systems)—to address emerging security and safety challenges in frontier models and multi-agent AI systems. It focuses on threats such as model provenance, supply-chain attacks, inter-agent exploitation, prompt injection, and reasoning-time interventions, fostering collaboration across adversarial ML, systems security, and agentic AI.

Official CFP Back to deadlines Verified September 9, 2026

Key deadlines

Verified September 9, 2026

Full paper

July 1, 2026

AoE

Conference timeline

Submission and decisions

Full paperKey deadline

July 1, 2026 · AoE

Paper fit

Contribution paths

A strong submission should clearly identify its contribution and evaluate it appropriately.

Full papers

Submissions of up to 6 pages (excluding references and supplementary material), conforming to the COLM template and submitted via OpenReview. Accepted papers are non-archival and may be presented as either a spotlight talk or a poster.

Research areas in scope

01

Topics

Adversarial machine learning for foundation modelsMulti-agent AI security and compositional threatsInter-agent attacks and trust exploitationPrompt injection and indirect prompt attacksMemory poisoning and RAG securityTool-use and API exploitationModel provenance, fingerprinting, and watermarkingModel stealing and supply-chain attacksBackdoors and poisoning in frontier model pipelinesAdversarial attacks on reasoning models and chain-of-thoughtTest-time adaptation and intervention securityRobustness of agentic and embodied AI systemsSecurity for multimodal models and VLAsAlignment and safety evaluation for frontier systemsTheoretical foundations of robustness, controllability, and identifiabilityGovernance, auditing, and accountability for AI agentsBenchmarks and red-teaming frameworks for agentic systems

Policies worth checking twice

  • Submissions must be up to 6 pages (excluding references and supplementary material).
  • Submissions must conform to the COLM template.
  • Submissions must be made via OpenReview.
  • Accepted papers are non-archival (will not appear in formal conference proceedings).
  • Concurrent submissions to other venues are allowed, but authors are responsible for ensuring compliance with those venues' policies.
  • Accepted papers will be allocated either a spotlight talk or a poster presentation.
  • The AdvML Rising Star Award requires a CV, a 2-page research statement, a 5-minute video, and two letters of recommendation submitted by specified deadlines.

Official sources

Compiled from the official call for papers. The organizers’ pages remain authoritative.

Last verified September 9, 2026

AdvML-Frontiers-CoTMA 2026: deadlines, venue, and submission guide | COREXA