AI Safety Indonesia

AI Safety Indonesia (aisafety.id) is a Jakarta-based community organization whose mission is to build "a grassroot community of AI Safety guardians" ensuring "our values and spirit of gotong royong are reflected in the AI technologies we use." Founded in 2025, the organization runs two programs — an AI Safety Professional Training Program developed in partnership with the Stanford Center for AI Safety, and a student-focused Outreach Program — and positions itself as the primary field-building node for AI safety education in the world's fourth-most-populous country and the largest nation in Southeast Asia.

Founding and context

The organization's founding framing explicitly addresses the distinctiveness of Indonesia's position: a country of 280 million people, the world's largest Muslim-majority nation, the largest economy in Southeast Asia, with deep linguistic and cultural diversity spanning hundreds of languages and ethnic groups, and a digital-infrastructure profile that makes it a major consumer and deployment target for AI systems developed elsewhere. AI Safety Indonesia's organizing concept — gotong royong, the Indonesian cultural principle of communal mutual assistance — positions AI safety as a project of cultural self-determination rather than import: Indonesian communities shaping the AI deployed among them on Indonesian terms, rather than receiving Western safety frameworks wholesale.

The organization's public activities in early 2025 reflect a deliberately multi-stakeholder model. A May 3, 2025 webinar commemorating National Education Day brought together speakers from the Stanford Center for AI Safety, Satu Data Indonesia at Bappenas (the national development planning agency's open data body), MindSTEM Indonesia, and the University of Pittsburgh — bridging the academic AI-safety pipeline, Indonesian government data infrastructure, and the Indonesian diaspora in US technical and scientific communities. The collaboration with Satu Data Indonesia grounds the organization's work in Indonesia's public data governance apparatus, situating AI safety not as purely technical but as embedded in how state institutions handle data at scale.

Programs

AI Safety Professional Training Program. The organization's flagship program is a three-course sequence developed in collaboration with the Stanford Center for AI Safety, with official certification from Stanford Center for AI Safety awarded on completing each course:

• AI Safety Fundamentals — a 6-day intensive (or 6-week distributed) fully remote course covering AI safety terminology, risk framing, alignment fundamentals, and the governance toolbox.
• AI Safety Principles in 4D — a 3-day in-person intensive in Jakarta applying safety engineering across the four stages of the AI lifecycle: design, development, deployment, and decommissioning. Covers risk assessment, adversarial defense, and implementation strategies.
• AI Safety Professional — a 6-day intensive at the Stanford campus, including site visits to leading technology companies. The capstone credential of the sequence.

Batch 02 opened applications April 7–25, 2025 with the program starting May 5–9, 2025 — indicating Batch 01 had already run. The program is explicitly aimed at professionals and organizations working with AI systems, rather than students entering the field: the target is practitioners who need AI-safety literacy anchored in Indonesian deployment contexts. Yan Pratama Akhra serves as Program Manager.

Outreach Program. A student-oriented complement targeting high school and undergraduate students through workshops, competitions, and mentorship — building foundational AI-safety awareness upstream of the professional pipeline. Annual AI safety competitions and a symposium are planned.

The organization also maintains a newsletter at aisafetyindonesia.substack.com covering topics such as culturally-aware AI red teaming for the Indonesian market, and hosts events through Luma (seven events hosted as of mid-2025).

Place in the movement

AI Safety Indonesia is the corpus's first entry for an AI safety community-building organization in Indonesia — the largest country in Southeast Asia and one of the world's major digital markets. It sits in the insider-community pattern alongside AI Safety Diliman, AI Safety India, Berlin AI Safety, and AI Safety 東京: community groups building the local practitioner pipeline for AI-safety work with EA-adjacent and field-building support connections.

What distinguishes AI Safety Indonesia's approach is the professional-certification emphasis and the cultural localization framing. Where AI Safety Diliman centers the governance perspective of a consumer state and AI Safety India leads with a curriculum moving students from fundamentals to research, AI Safety Indonesia's flagship program is aimed squarely at working professionals and explicitly addresses Indonesian deployment contexts — algorithmic bias in multicultural settings, data privacy, public safety — rather than leading with technical alignment research. The gotong royong framing is the clearest cultural localization signal in any entry in this corpus's AI-safety local-group layer.

The regional network the organization has established — partnerships with AI Safety India, AI Safety Asia, and AI Safety Korea, alongside the Stanford Center for AI Safety collaboration — situates it within the emerging non-Western AI-safety infrastructure layer whose defining question is how countries outside the frontier developer community can build domestic capacity to govern AI systems that are largely produced elsewhere. Indonesia's scale (fourth most populous country globally), its governance complexity (a decentralized archipelago state spanning 17,000 islands), and its exposure as a large consumer market for frontier AI systems all make the question of Indonesian AI-safety capacity distinctly consequential for the broader regional movement.