Code of Conduct

Preamble

COAI builds a community for open, responsible, collaborative research on LLMs. This Code of Conduct applies to all interactions with COAI infrastructure, events, online communities, and publications. Our aim is to create a respectful, safe, and inclusive environment for everyone.

1. Core Principles

We commit to:

  • Scientific integrity and transparency
  • Open knowledge sharing and reproducibility
  • Valuing diverse backgrounds and viewpoints
  • Supporting collaboration over competition
  • Fostering thoughtful, critical engagement with LLMs and their societal impact

2. Expected Behavior

Participants should:

  • Treat others with professionalism, kindness, and respect
  • Communicate constructively, especially in disagreement
  • Provide and receive feedback in good faith
  • Contribute collaboratively during events and in research
  • Credit contributions fairly and acknowledge sources
  • Follow applicable laws, licenses, and research ethics guidelines

3. Unacceptable Behavior

The following conduct is prohibited:

  • Harassment, discrimination, or targeted exclusion
  • Degrading, threatening, or aggressive language or behavior
  • Sabotaging experiments, sessions, or discussions
  • Sharing private data or content without consent
  • Misusing shared infrastructure or bypassing access controls
  • Prompting or deploying LLMs in ways intended to generate harmful content

4. Responsible Use of Infrastructure

Users must:

  • Use only assigned credentials and not share access
  • Respect fair-use limits and access conditions
  • Report bugs to appropriate contacts
  • Avoid unauthorized data extraction or commercial misuse
  • Accept responsibility for inputs and outputs

5. Events & Community Spaces

Participants should:

  • Engage constructively with fellow members
  • Respect organizers, facilitators, and mentors
  • Observe time limits and moderation rules
  • Avoid trolling, grandstanding, or dominating discussions
  • Use public channels for event-related help

6. Attribution & Publication

Projects from COAI-supported activities require:

  • Acknowledging COAI’s events and infrastructure used
  • Following citation guidelines from COAI or providers
  • Respecting open-source licensing terms

7. Enforcement & Reporting

Violations may be reported to: edif@coairesearch.org

Consequences may include:

  • Warnings (informal or formal)
  • Temporary or permanent access revocation
  • Reporting to affiliated institutions for severe misconduct

All reports are handled confidentially and in good faith, while protecting community integrity.

8. Continuous Feedback

This Code of Conduct is a living document that evolves as COAI grows. We welcome community feedback and suggestions for improvement.

9. Use of Generative AI in Publications

COAI recognizes that generative AI tools, including large language models (LLMs), are integral to research in our community—both as subjects of study and as aids in the research process. The following guidelines balance openness with accountability.

9.1 For Authors

General Principle: Authors are welcome to use any tools that help produce high-quality research, including LLMs. Authors bear full responsibility for all content—text, figures, code, and references.

Disclosure Requirements:

  • No disclosure required for grammar and spelling corrections, style improvements, translation assistance, or code formatting/debugging
  • Disclosure required (in Methods or Acknowledgments) when LLMs are used as an important, original, or non-standard component of the research methodology itself
  • When disclosing, include: tool name/version, purpose of use, and how outputs were verified

Prohibited:

  • Listing AI tools as authors or co-authors
  • Fabricating data, references, or citations
  • Submitting unverified AI-generated content (authors must check all outputs for accuracy)

9.2 For Reviewers

COAI adopts a dual-policy system that respects both author preferences and reviewer workflows.

Policy A (Restrictive):

  • Submissions may NOT be uploaded to any LLM tools
  • LLMs may only be used to understand general concepts (without sharing submission content) and to polish the reviewer’s own writing

Policy B (Permissive):

  • Submissions may be uploaded to privacy-compliant LLMs (enterprise APIs with training opt-out, self-hosted models, or services with explicit data protection)
  • LLMs may be used to understand the paper and related work
  • LLMs may NOT be used to assess quality, identify strengths/weaknesses, draft the review structure, or write the review itself

How it works:

  • Authors select their preferred policy (A or B) at submission
  • Reviewers indicate which policy they are willing to follow
  • Papers are matched accordingly

Under both policies:

  • The scientific assessment must be the reviewer’s own judgment
  • Confidentiality of submissions must be maintained
  • Reviewers are fully responsible for their review content

9.3 For Editors

  • Editors must not use LLMs to draft decision letters or summarize unpublished research
  • Confidentiality requirements apply as for reviewers

9.4 Accountability

  • Human authors and reviewers bear full responsibility for accuracy and integrity
  • “Hallucinated” content (fabricated references, false claims) is treated as research misconduct
  • Violations may result in consequences as outlined in Section 7 (Enforcement & Reporting)

This policy recognizes that norms around AI use in research are evolving rapidly. We welcome community feedback and will revisit these guidelines as practices develop. Last updated: February 2026