AI Safety Türkiye

AI Safety Türkiye (aisafetyturkiye.org) is Türkiye's first civil-society AI safety community, co-founded in 2023 by Bengüsu Özcan and Berke Çelik as an independent, volunteer-sustained initiative. Its mandate is to "address AI's near and far-term risks together" — near-term societal impacts alongside longer-horizon alignment and existential risk concerns — and to ensure that AI technologies benefit humanity as they transform politics, economy, and society in the 21st century, with a deliberate framing of "global contributions from a Turkish perspective." Operating primarily from Istanbul, with a growing presence in Ankara, AI Safety Türkiye produces Turkish-language content, runs structured educational programmes, and participates in international advocacy networks. It is the corpus's first entry for a civil-society AI safety organisation in Türkiye, and the region's second such entry alongside AI Safety UAE.

Founding and structure

AI Safety Türkiye was founded in 2023 and as of October 2024 remains unregistered — an independent community with no formal legal-entity status. This structure is common for early-phase EA-adjacent AI safety groups globally: it keeps operating overhead low while the community tests programme formats and builds a practitioner base.

Co-founder Bengüsu Özcan holds a BS in Industrial Engineering and Psychology from Sabancı University and an MA in Quantitative Methods in Social Sciences from Columbia University, where her thesis addressed user perceptions of recommendation algorithms. She serves as an Advanced AI Researcher at the Centre for Future Generations (CFG), a Brussels-based research and policy institution, and facilitates AI governance courses for both BlueDot Impact and AI Safety Türkiye. Her policy work on compute governance has been featured in the context of the G7 summit. She is also a published author of two Turkish-language novels, Maskeli Kedi and Antika Adam (English: Ancient Man), and a children's book, Kendi Yolunda, designed as a nonprofit initiative addressing gender bias in career aspirations.

Co-founder Berke Çelik holds a degree in Philosophy from Boğaziçi University — one of Türkiye's leading research universities — and serves as director of the development programme at the Global Policy Research Group, with a research focus on AI policy and decision theory in developing countries. Like Özcan, Çelik is also affiliated with the Centre for Future Generations. The two co-founders' shared CFG affiliation gives AI Safety Türkiye an unusually direct channel into Brussels-based AI governance policy work for a volunteer-run community group. The broader volunteer team includes Sayhan Yalvaçer, Alparslan Bayrak, and İrem Sena Karakoç.

Programmes — Turkish-language content and structured education

AI Safety Türkiye produces several recurring offerings tailored to Turkish-language audiences, filling a gap in a field whose primary literature is almost entirely English:

• Blog — in-depth Turkish-language analysis spanning engineering, policy, law, and philosophy. Pieces include system card reviews and comparative policy analysis, among them a comparison of the EU AI Act and Türkiye's own AI legislation proposal, positioning the blog as the primary Turkish-language venue for substantive AI safety writing.
• Newsletter — a monthly digest covering AI safety developments, with 134+ subscribers as of 2024 — one of the few regular AI safety periodicals in Turkish.
• Introduction to AI Governance — a structured five-week reading and discussion cohort. The first cohort launched in March 2024; the second ran in October 2024, hosting Dutch AI governance researcher Jelle Donders as a guest participant and drawing applications at nearly twice the available capacity. The programme is designed for Turkish practitioners and students seeking a structured entry into AI governance thinking.

Outreach — university engagement and community meetups

AI Safety Türkiye has built a sustained university outreach operation since its founding. The group presented an introduction to AI safety in August 2023 in collaboration with Koç University's Effective Altruism Club in Istanbul, one of Türkiye's leading private research universities and a hub of EA-adjacent activity. The founders subsequently ran booths and discussions at the Bilkent University Philosophy Festival in both September 2024 and September 2025 — Bilkent, located in Ankara, is one of Türkiye's leading research universities — engaging students through polls and conversation about AI opportunities, risks, and safety-research career paths.

Community meetups have run in Istanbul from August 2023, with the Kadıköy Istanbul Kitapçısı location serving as a gathering point for newcomers and regular members. In October 2024 the group held its first in-person meetup in Ankara, marking a geographic expansion into the capital. In August 2025, AI Safety Türkiye ran a joint session with EA Koç University. In October 2025, co-founder Bengüsu Özcan delivered an AI safety session at EA Summit: Istanbul 2025 (Kolektif House Levent, organised by Effective Altruism Turkey), a Türkiye-wide EA gathering conducted entirely in Turkish.

International advocacy — Red Lines and Athens Roundtable

Beyond community-building, AI Safety Türkiye participates in international AI policy advocacy. The organisation endorsed the Global Call for AI Red Lines, a campaign launched on September 25, 2025, co-organised by The Future Society, the French Centre for AI Safety (CeSIA), and the Center for Human-Compatible AI (CHAI) at UC Berkeley. The call asks governments to establish binding red lines for the most dangerous AI use cases; it was signed by more than 200 individuals — including former heads of state, Nobel and Turing laureates, and AI researchers — and more than 70 organisations.

On December 4, 2025, AI Safety Türkiye participated in the Seventh Athens Roundtable on AI and the Rule of Law, convened in London by The Future Society in partnership with OECD and UNESCO. The Roundtable gathered 140 global leaders and experts; a closed-door session brought together 76 experts to discuss operational AI red lines, enforceable technical thresholds, verification mechanisms, and multilateral standards grounded in existing international law. AI Safety Türkiye's stated emphasis at the Roundtable was on "binding AI red lines and collective action on shared risks."

In January–February 2026, AI Safety Türkiye co-hosted the Ankara hub of the Next Gen AI Forum 2026 with Encode, a technology governance and education community, running panels, networking, and workshops at Unite Ortak Mekan in Ankara. The collaboration marked AI Safety Türkiye's first jointly organised event with a wider technology-governance partner in the capital.

Place in the movement

AI Safety Türkiye is the corpus's first AI safety community group entry for Türkiye, and the second for the Turkey/Middle East region alongside AI Safety UAE. What distinguishes AI Safety Türkiye from its regional counterpart is the emphasis on Turkish-language content production — the blog and newsletter are deliberately in Turkish, filling a gap for audiences not well-served by the English-dominant global AI safety canon — and on university-anchored outreach, consistent with a founding team rooted in research and policy rather than enterprise services.

Among the corpus's AI safety community groups globally, AI Safety Türkiye most closely resembles the early-phase AI Safety 東京 and Berlin AI Safety models — volunteer-led, community-building organisations that develop practitioners, produce field-accessible content, and engage student audiences — rather than the more institutionalised models of the Singapore AI Safety Hub (SASH) or the hybrid commercial/community model of AI Safety UAE. The co-founders' dual roles at the Centre for Future Generations give the group a more direct policy-research channel than most volunteer-led meetup groups at this stage — an access that reflects the founding team's professional roots rather than the community's size.