Are you passionate about AI safety and ready to dive into cutting-edge research? The Anthropic Fellows Program 2026 offers a chance for technical talent to gain funding, mentorship, and hands-on experience in AI development.
This program supports promising researchers and engineers, no matter their past experience level. Fellows focus on empirical projects using tools like open-source models and public APIs. The goal is to create public outputs, such as research papers. In past groups, more than 80% of participants published papers.
Program Workstreams
Anthropic is growing the program across different teams. Skills often overlap, so applicants get considered for multiple areas. You can list preferences, but some streams have extra evaluation steps.
Here are the main workstreams:
- AI Safety Fellows: Work on making AI systems reliable and aligned with human values.
- AI Security Fellows: Tackle risks like hacking or misuse of AI models.
- ML Systems & Performance Fellows: Improve machine learning tools for speed and efficiency.
- Reinforcement Learning Fellows: Build AI that learns from trial and error, like in games or robotics.
- Economics & Societal Impacts Fellows: Study how AI affects jobs, markets, and communities.
Key Benefits
The program lasts 4 months of full-time work. You get direct guidance from Anthropic researchers. Fellows share a workspace in Berkeley, California, or London, UK.
Other perks include:
- A weekly stipend: $3,850 USD, £2,310 GBP, or C$4,300 CAD, plus country-specific benefits.
- Up to $15,000 per month for computing power and research costs.
- Links to the wider AI safety and security network.
Eligibility Requirements
To join, you need to be fluent in Python and free for full-time work. Strong applicants often have skills tied to a workstream, like economics, social sciences, or cybersecurity.
You might fit well if you:
- Care about safe and helpful AI for everyone.
- Want to shift into hands-on AI research and consider a job at Anthropic later.
- Hold a solid background in computer science, math, or physics.
- Enjoy quick teamwork and clear idea-sharing.
How to Apply
Anthropic reviews applications year-round in batches. This round targets starts from July 2026 onward. The next group begins July 20, 2026.
Submit by April 26, 2026, for the best shot. Apply here. Check full details at the Anthropic Fellows page.
Frequently Asked Questions
What is the Anthropic Fellows Program?
It’s a 4-month program for researchers and engineers to work on AI safety projects with funding, mentorship, and hands-on experience using open-source tools.
What workstreams are available?
Options include AI Safety, AI Security, ML Systems & Performance, Reinforcement Learning, and Economics & Societal Impacts.
What benefits do fellows get?
Fellows receive a weekly stipend, up to $15,000 monthly for computing, workspace in Berkeley or London, and connections in the AI network.
How do I apply for the 2026 program?
Submit by April 26, 2026, via the application link for starts from July 2026; applications are reviewed in batches year-round.