IASER: Building Capacity for Safe, Ethical, and Responsible AI
IASER began in July 2026 to bring Muslim researchers, scholars, and technology professionals together around AI safety. More than 100 people have joined since then. We have made progress on AI evaluations, built tools including Ansari, and started developing practical guidance and training.
The challenge now is keeping the work moving. Most of our members are volunteers with jobs and other commitments. We have research projects underway, but not enough people with dedicated time to coordinate them, follow up on tasks, or maintain what we have built.
We are seeking $100,000–$500,000 over the next 12 months. Our first priority is to support the people already doing the work, finish what we have started, and give IASER a stronger foundation.
Goals and approach
We do not have a shortage of ideas or people willing to help. What we lack is the capacity to organise that work well.
Our first priority is project coordination. With more than 100 members, we need better ways to match volunteers to tasks, clarify responsibilities, and carry projects through to completion. We also need to be realistic about how many projects we can take on.
The second priority is our existing research:
JaleesBench and FaithfulBench examine how AI models respond to religious questions and whether they hold steady under pressure. JaleesBench tests whether an AI assistant acts as good spiritual company, using 140 two-turn scenarios and measuring how models respond across adversarial pressures and framings.[arxiv]
FaithfulBench scores AI counsel across religious traditions by how well it adheres to a user’s professed faith, testing models with and without information about the user’s tradition and with a tradition-specific counselling guide.[arxiv]
JaleesWeights explores model post-training.
Ansari is a free Islamic AI assistant.
We want to continue these projects, improve the evaluations, and publish findings that others can examine and build on. That requires computing resources, technical work, review, and maintenance.
We are also developing an Islamic Ethics Guidebook for Tech Professionals. Many Muslims working in technology face difficult questions about the systems they help build, especially around surveillance, discrimination, and other potential harms. We are gathering evidence of these challenges and developing guidance for real situations. We are also collaborating with Yaqeen Institute on an Islamic Ethics Framework for AI.
Education is another priority. Dr Waleed Kadous and collaborators have developed and delivered practical AI workshops for imams. In a pilot survey of 84 imams, more than half said they used AI in their religious work, but fewer than half felt equipped to answer questions about it.
We want to improve this training and make it available more widely, with particular attention to misinformation, fabricated references, and knowing when AI outputs need checking.
We do not expect to do everything at once. In the first two months, we will review our projects, agree on priorities, and assign responsibilities based on the funding available.
How we will measure success
A year from now, IASER should be finishing more of the work it starts and relying less on a small group of volunteers.
We will track completed projects, research publications, volunteer contributions, and use of our resources. For research, we will report what we tested and found, including results that do not support our expectations.
For the ethics guidebook and training, we will ask users whether the material meets their needs and what should change.
We will review finances and project progress every quarter. After six months, we will decide which activities to continue and which need to change. At the end of the year, we will publish a report covering what we achieved, what did not work, and how the funding was spent.
Use of funds
We are requesting $100,000–$500,000 for 12 months.
Most of the budget would go to people: dedicated time for research, engineering, volunteer coordination, and project management. The remainder would cover research costs, maintenance of tools such as Ansari, educational resources, training, and essential operating expenses.
We will adjust the scope according to how much we raise:
$100,000: project coordination, essential services, and a small number of priority projects.
$250,000: more research and community activity.
$500,000: the wider programme, with dedicated staff or contractors where needed.
These are estimates. We will not commit spending until we have confirmed actual costs and responsibilities.
Team and track record
IASER is new, but our team brings experience from AI research, technology companies, education, Islamic scholarship, and nonprofit organisations.
Our Steering Committee includes:
Dr Waleed Kadous, who has worked in AI for more than 25 years, including senior roles at Google, Uber, and Canva. He developed Ansari and the AI for Imams course and has contributed to our model evaluation research.
Dr Mohammed Ansari, a senior applied scientist at Amazon with a PhD in Computer Sciences from the University of Wisconsin–Madison.
Dr Agnes Chong, who has worked in nonprofit leadership and philanthropy, including roles at the American Muslim Community Foundation and CAIR.
Bara Elhag, who brings experience in technology sales, business development, and consulting.
Abdus-Sami Hoda, who works in education and technology leadership and is involved with Ummatics Institute and Mubeen Institute.
Hassan Kazi, who has a background in management consulting at Deloitte and Accenture and serves on the board of Justice For All Canada.
Our Islamic Advisory Board includes Dr Mohamed AbuTaleb, Vice President of Research at Yaqeen Institute, and Shaykh Umer Khan, a member of the Fiqh Council of North America and a technology executive.
These are not full-time positions at IASER.
We also have work people can examine today:
We have also organised public discussions and presentations on AI and its implications for Muslim communities.
Some of this work began before IASER formally launched. We are now bringing these efforts together and building an organisation that can support them properly.
More information about our team is available at iaser.ai/about.
Risks and contingency
The biggest risk is spreading ourselves too thin.
Much of our work depends on people contributing in their spare time. Projects often begin with enthusiasm, then slow down when other commitments take priority.
Funding would increase our capacity, but we would still need to make careful choices about what to pursue. We could underestimate costs, struggle to find the right people, or find that something we developed is less useful than expected.
We plan to review projects regularly and stop work that is not progressing. If we do not raise enough money, we will scale back rather than try to deliver the full programme.
If the wider plan does not succeed, the most likely outcome is that some research and community resources take longer to complete, and IASER remains dependent on volunteer time. We will report problems as they arise and account for how funding has been used.
Funding to date
IASER launched in July 2026 and is still establishing its fundraising arrangements.
Most of our work so far has been carried out by volunteers and founding members, using their own time and resources. We have grown to more than 100 members and are now approaching potential funders to help sustain and expand this work.
We are developing funding proposals and donor relationships, but we will not count applications or informal discussions as money raised. We will update our fundraising figures as commitments are confirmed.
This campaign is an important step towards giving IASER the people and resources it needs to move beyond volunteer-led work and become a more sustainable organisation.