The race to shape safe and responsible artificial intelligence has entered a new phase as the Machine Learning Alignment & Theory Scholars programme, widely known as MATS, opens applications for its Autumn 2026 fellowship cohort.
- The race to shape safe and responsible artificial intelligence has entered a new phase as the Machine Learning Alignment &...
- The fully funded fellowship is drawing attention from aspiring AI researchers, policy experts, and cybersecurity professionals seeking to address the...
- The organisation noted that many former fellows now work in AI safety and governance roles across research institutions and technology...
- Organisers said the support package is intended to allow fellows to focus fully on research and collaboration during the programme.
Keep reading for the full breakdown on AI alignment — everything you need to know is covered below.
The fully funded fellowship is drawing attention from aspiring AI researchers, policy experts, and cybersecurity professionals seeking to address the growing risks posed by advanced artificial intelligence systems. Applications close on 7 June 2026.
Global Mentors Back AI Safety Fellowship
The programme connects fellows with mentors from leading AI organisations, including OpenAI, Anthropic, Google DeepMind, Redwood Research, and the Alignment Research Centre. Participants will work on research projects focused on AI alignment, governance, transparency, systems security, and biosecurity.
MATS said the fellowship aims to strengthen global efforts to manage the risks associated with increasingly capable AI systems. The organisation noted that many former fellows now work in AI safety and governance roles across research institutions and technology companies. Over the past four years, alumni have reportedly contributed to around 190 research publications with more than 12,000 citations.
The programme is expected to begin on 28 September 2026 and will provide fellows with research support, mentorship, and access to an international AI safety network.
Fellows to Receive Funding and Visa Support
Selected participants will receive a stipend of $12,500, along with up to $20,000 in computing support for research activities. The fellowship will also cover housing, meals, and travel expenses. International participants who require relocation support may receive J-1 visa sponsorship.
Organisers said the support package is intended to allow fellows to focus fully on research and collaboration during the programme.
Applicants can apply to one or more research tracks depending on their interests and experience. Available tracks include empirical AI alignment, theory, policy and governance, strategy and forecasting, systems security, founding and field building, and biosecurity. Each track contains specialised, mentor-led streams.
Multi-Stage Selection Process Announced
The fellowship application process will unfold in three stages. The first stage requires candidates to submit educational and professional details, short written responses, and references.
Shortlisted applicants will then move to central evaluations tailored to specific tracks. These assessments may include technical implementation tasks, analytical reasoning exercises, and writing evaluations.
The final stage will involve mentor stream applications, in which candidates rank their preferred streams and complete interviews, essays, research proposals, or work tests, depending on the selected pathway. Final decisions are expected in late July or early August 2026.
As governments and technology firms face mounting pressure to strengthen AI oversight, programmes such as MATS are becoming an increasingly important pathway for researchers seeking to shape the future of safe and accountable artificial intelligence.
Interested applicants should apply here for the programme.