Approved by the AOS Council on 3 August 2026
Background and Purpose
Artificial Intelligence (AI) is an emerging tool that has many potential positive uses for associations, from tailoring content to member interests, to supporting data-driven decision making, to automating tasks to free staff and volunteer time for strategy and oversight. However, AI requires careful human oversight, and it is critical to adhere to guidelines to ensure data privacy, support accountability, mitigate biases, promote transparency, and protect intellectual property. Doing so will allow the AOS to use AI as a valuable tool in ways that keep confidential data safe, ensure that we adhere to applicable laws and copyright protections, and maintain our members’ trust.
The AI landscape changes quickly, with tools, uses, and laws continuing to evolve. AOS is evaluating how volunteers and staff currently use AI in our programs, operations, and governance, and how members and applicants use AI in their engagement with the AOS. We are also evaluating the possibilities for how AI can be used in the future to support programs, members, volunteers, and staff. As we go through this process of understanding current AI use, anticipating future use, and developing comprehensive guidelines that consider the many factors in AI opportunities and challenges, the AOS needs some interim guidelines to ensure that AI use and experimentation within the AOS does not inadvertently put our members, our programs, or our credibility and trustworthiness at risk.
These proposed guidelines are meant as an initial framework for the AOS and should be revisited and updated as our more thorough evaluation of opportunities and risks yields more specific information on guidelines that will balance AI innovation with risk mitigation.
Scope
This policy applies to all employees, volunteers, and contractors using AI tools for association work. This includes but is not limited to:
Public AI tools (e.g., ChatGPT, Claude)
- AI features within existing software
- Association-specific AI solutions
- AI-powered automation tools
Policy
DATA PRIVACY AND PROTECTION
- Any AI use within the AOS must comply with all applicable laws, including data privacy laws.
- Sensitive or confidential personal, professional, or financial information should never be input into any free AI tool. This includes:
- Protected member demographic information (contact information, other personally identifiable individual member data)
- Personal financial information (banking, credit card, etc.)
- Social Security numbers/government issued ID numbers
- Personnel information
- Confidential contracts/legal documents
- Council/executive committee minutes or notes
- Financial statements/budgets
- Unpublished abstracts, articles, or other scientific content
- Grant applications
- Private correspondence
- Confidential AOS governance or program information should not be input into any AI tool that is not paid for and authorized by the AOS. This includes:
- External (non-AOS) AI notetakers are not allowed at Council, Executive Committee, or other AOS governance meetings
- External (non-AOS) AI notetakers are not allowed at program meetings where sensitive information is discussed (e.g. grant or award meetings where individuals or individual applications are discussed)
- AOS Council minutes, non-public financial statements, etc should not be input into external AI tools (that is, tools that are not approved and paid for by the AOS)
- Use caution when entering sensitive data into any paid AI tool, including tools paid for, developed, and/or owned by the AOS.
- Data privacy should be prioritized, and data protection regulations must be adhered to.
- If AOS volunteers or staff use a paid AI tool, sensitive data must be anonymized.
- Information and data that are generally safe to use with AI
- Public information
- Anonymized aggregate data
- Published materials
- Certain information should NEVER be entered into any AI tool, paid or unpaid, that could expose an individual to a risk of identity theft or harm (e.g. banking or credit card information, social security numbers, etc)
- AOS members, volunteers, and staff who use AI for any work associated with AOS should only use reputable tools.
Transparency
- AOS staff and volunteers must disclose AI use when:
- Publishing member-facing or public-facing content created with AI
- Distributing AI-assisted educational, training, or guidance materials, or creating resources with significant AI input
- Sharing AI-generated analysis or insights
- Using AI to substantially edit member content
Bias Mitigation
Using AI tools can inadvertently perpetuate biases in selection processes. For example, if individuals with certain characteristics (e.g. gender, country of origin, publications primarily in non-English journals) were less likely to be nominated to be an Elective Member or receive an achievement award historically, AI algorithms might downgrade individuals with those characteristics if asked to develop a list of eligible members to consider moving forward. If AOS volunteers, members, or staff use AI to evaluate applications, eligibility, or to identify members who might take on certain roles, it is important to be aware of historical biases and how they might influence AI algorithms, and take steps to overcome these biases.
Accountability
- For any analysis of data– either internal AOS data or data used in AOS publications, presentations, or grant applications– AOS AI users must understand how to replicate the analysis and check the results for accuracy and to evaluate potential bias (see above). All individuals using AI within the AOS or in conjunction with presentation, publication, or application in/to AOS venues are responsible for the output of their AI tools, which must adhere to the same or higher standards of trustworthiness, validity, credibility, and repeatability of other AOS efforts. For all AI-generated content, AOS users must:
- Verify factual accuracy
- Check for appropriate tone
- Ensure policy compliance
- Confirm no confidential data
- Review for completeness
- Add required disclaimers
- AOS must respect others’ intellectual property (IP). AI users within the AOS are responsible for ensuring that any AI-generated product does not infringe on intellectual property rights.
- Any AOS content creators (staff, volunteers, etc) must obtain permission to enter member- or attendee-generated content into any AI tool to generate derivative works (e.g. summaries, blogs, etc) with a written release. The original content creator will have the opportunity to review and approve any derivative products before their release.
- AI does not replace professional judgment or expertise. Any products that are generated entirely or partially by AI must be reviewed by one or more relevant experts (AOS staff members, volunteers, or others) to ensure accuracy and validity, before they are distributed.
Policy Sources and Attributions
Many organizations are developing policies around the use of AI/LLMs, and several organizations offer guidance or templates for non-profit or association AI policies in this realm. The AOS draft policy reflects consideration of the following policies, templates, or sources.
Sidecar AI is an AI resource specifically for associations. Two of their blogs, AI policy development: Balancing innovation and data protection in associations and Top 6 AI guidelines for associations to follow, helped guide the structure and high-level content included in this policy. Sidecar AI also has an AI policy template that provided some specific language that was taken from or loosely adapted in the Scope, Transparency, and Accountability sections above. Thrive, a firm that provides AI, cybersecurity, and other IT services, also has an AI policy template that was used for reference.
An AI Usage Policy from the Society for Scholarly Publishing (June 2025) and the staff AI policy from INFORMs (March 2026) were also reviewed for context and content as the AOS policy was being drafted.
This AOS draft policy also reflects guidance and considerations from AI-specific discussions at the Council for Engineering and Scientific Society Executives ACCESSE Annual Meeting (June 2025) and CEO meeting (January 2026), and from the Oxford University Press Journals Executive Summit (April 2026).
Additional Note: If the AOS adopts a policy that limits certain AI uses to tools paid for by the AOS, the AOS will need to provide a paid AI tool for staff and volunteers. Evaluating AI tools that will fit the AOS’s needs is a task staff are working on.