This was extracted (@ 2026-09-18 21:10) from a list of minutes
which have been approved by the Board.
Please Note
The Board typically approves the minutes of the previous meeting at the
beginning of every Board meeting; therefore, the list below does not
normally contain details from the minutes of the most recent Board meeting.
ASF Members may have access to a
private draft
WARNING: these pages may omit some original contents of the minutes.
# Responsible AI Initiative Board Reports ## Report 1: August 2026 **Description** The ASF Responsible AI Initiative (RAI) is a program to support the open source technologies that underpin modern artificial intelligence, and to help Apache communities adopt AI in a way that is open, transparent, and aligned with the ASF's values. The initiative is grounded in the ASF's guidelines for the responsible use of AI — human oversight, licensing integrity, security, and documentation — core principles rooted in the Foundation's longstanding philosophy of community over code. To fulfill its charter, the RAI is undertaking efforts to facilitate: 1. Access to AI models and tooling 2. Ecosystem support for Projects 3. Community engagement and outreach **Status** Status on the above activities includes supporting the Security Scanning & Triage collaboration between the ASF Security team and the Tooling team. More than 75 PMCs have enlisted: some delays from one of our frontier model donors are now resolved and initial scans are underway. I participated on calls with Infrastructure, Tooling, and Fundraising to coordinate targeted donations, prioritize token usage, and connect with Initiative backers. We are planning to conduct a survey of PMCs and operations teams on their use of AI to have a better idea on the state of AI activity and tooling within the ASF. We are touching base with the folks on Apache Magpie and coordinating various projects-in-development, including Low-Rank Adaptation (LoRA) adapters and a dynamic routing layer, a TLP proposal being prepared for Apache Sourcelume — an AI training data provenance infrastructure, as well as a proposal preparing to be submitted to the Apache Incubator or for new Project (TBD). Initial work has begun on LLMAO (LLM for Apache.Org), a Foundation-wide way for committers, projects, and communities to reach sanctioned inference and related AI capabilities without every PMC holding its own provider keys. The goal is one governed chokepoint for attribution, budgets, and policy for work in an AI-capable ASF. This project is to provide gateway access through LiteLLM to hosted models including frontier and cloud (local) implementations. We have opened the floor for ideas regarding budget, and worked on infrastructure for RAI: our initial site is up at https://rai.apache.org/ . We plan to add a new logo for RAI and are narrowing down to a handful of options. We had a discussion with the Community Over Code/Asia panelists to review messaging, and shared a new graph of our email counts that shows the source of our pull requests (49% from unknown authors: possibly bots; need to look into this). As we're just officially getting started (our second week!) we welcome Member feedback and ideas at `discuss@rai.apache.org` and at `#rai-discuss` on Slack. We have currently put together an initial budget, in coordination with Tooling and Infra to obtain some initial numbers primarily for hosting private LLMs and token acquisition. The budget also includes travel for the VP to Glasgow/CoC surrounding events with potential for one or more presentations, Q&A, panel, BoF, and hackathon. The budget request is for $83,500.00 for the current fiscal year. A resolution to approve the budget will be submitted to the Board for consideration. **Upcoming** We aim to share results as we develop our tooling and processes. We are planning to attend Community Over Code in Glasgow, and look forward to sharing our progress with the greater community. **Work items** From our proposal for RAI, we have begun our initial work items: 1. Assess foundation-wide the current state of AI activity and tooling 2. Prioritize AI resources PMCs would like the Foundation to provide 3. Assemble a situation report with gap analysis 4. Prepare a strategy within 60 days 5. Maintain a rolling 60-day roadmap and plan 6. Construct budget updates for the Board's consideration 7. Collaborate with existing teams on policy We are assembling estimates for potential work streams to support budget plans. And we will be coordinating with existing teams this month to assess any policy needs.
