Who Verifies
the Agents?

Topics of Interest

We invite submissions across three core pillars, as well as topics at their intersection:

Pillar 1: Safety and Robustness of Verification

• Robust verifiers that prevent reward hacking and specification gaming
• Adversarial robustness of verifiers and red-teaming of evaluation harnesses
• Alignment-aware verification: ensuring verifiers remain faithful as agents evolve

Pillar 2: Environment-Grounded Verification and Simulators

• Faithful simulators as verification infrastructure for open-ended tasks
• Multi-agent and self-optimizing systems for environment-grounded evaluation
• Measuring agents in production: observability, monitoring, and runtime verification
• Evolutionary and search-based methods for environment-driven agent optimization
• Benchmarks and environment design that stress-test verification methods

Pillar 3: Diverse and Heterogeneous Verifiable Signals

• Composing heterogeneous signals (user experience, cost/latency, calibration, multimodality) into reliable verification metrics
• Beyond scalar rewards: holistic evaluation of agentic behavior
• Human-in-the-loop verification and human–AI collaborative evaluation
• Reflective and self-improving verification (agents that verify other agents)
• Automated agent design, prompt optimization, and scaffold search with verifiable feedback
• Formal verification of agent-generated artifacts, including the use of proof assistants and verification languages (e.g., Dafny, Rocq, and Lean)

Cross-Cutting Topics

• Verification for meta-agents and Agent4Agent systems (agents that design, optimize, or evaluate other agents)
• Self-evolving agents: stable improvement without collapse or reward hacking
• Evaluation of agent-generated designs vs. human-engineered systems
• Scalable oversight and verification for long-horizon, multi-step agent behavior
• Cognitive and neuroscience-inspired verification frameworks

Submission Guidelines

• Format: Papers should be between 4 to 9 pages (excluding references and appendices), using the NeurIPS 2026 template. We also welcome demo papers, which should be no more than 4 pages.
• Dual submission policy: We welcome work that is under review or has been recently published at other venues.
• Review: Reviews will be double blind. Authors of submitted papers may be asked to contribute reviews.
• Presentation: Accepted papers will be presented as posters; select papers will be chosen for oral presentations or lightning talks.
• Non-archival: The workshop is non-archival; accepted papers will be made available on OpenReview but do not constitute formal proceedings.

Submissions and reviewing are handled through OpenReview.

Call for Reviewers

We are assembling the program committee and welcome researchers and practitioners working on agents, evaluation, and verification.

Guidelines

• Scope: Evaluate submissions based on their contributions to the verification, evaluation, reliability, robustness, and safety of AI agents. We welcome work on topics including safety, simulators, heterogeneous signals, meta-agents, benchmarks, and evaluation methodologies.
• Reviewer Responsibilities: Provide constructive feedback on assigned submissions. Maintain confidentiality throughout the review process and declare any conflicts of interest.
• Area Chair (AC) Responsibilities: ACs are expected to oversee assigned reviews, ensure reviews are fair and constructive, and submit final recommendations of assigned submissions.

Sign up as a reviewer →

All times are local Sydney time (AEDT) on Saturday, December 12, 2026.

Reviewers

• Can I still volunteer to review? I accepted the invitation but see no assigned papers.
  Reviewer assignments for the 2026 cycle are finalized. We will keep your details on file in case emergency coverage or reassignment is needed. If you do have assignments, please complete them on time.
• I am a first-time reviewer. Should I still complete my reviews?
  Yes. Review to the best of your ability, and note explicitly where you feel uncertain; Area Chairs and Program Chairs will take that context into account.
• I have a conflict of interest, an AC/reviewer overlap, too many papers, or an assignment outside my expertise. What should I do?
  Email the organizers promptly with the affected submission(s) and the reason; we can remove or reassign papers. Please still complete your other reviews on time.
• A submission I am reviewing is not anonymized. Do I need to do anything?
  Flag it confidentially to the organizers; nothing further is needed on your side.
• Can I receive a letter confirming my reviewer or Area Chair service?
  We will not provide individual confirmation letters. Instead, reviewers and Area Chairs will be acknowledged by name on the workshop website.

Authors and submissions

• Are late submissions accepted?
  No.
• Can I update my paper, abstract, artifact link, citation, or author list after submission?
  Not at this stage. If your paper is accepted, you can make these updates during the camera-ready process.
• What should I do if I accidentally submitted a duplicate?
  Withdraw the unwanted copy using the Withdraw button in OpenReview, then confirm to the organizers which submission should remain.
• Is the NeurIPS Paper Checklist required?
  No. It is recommended but not required, and papers without one are reviewed normally.

Still need help?

Email verify-agents-workshop@googlegroups.com with your OpenReview profile, submission number(s), and a concise description of the issue.