The Anthropic Fellows Program is a structured research initiative created by Anthropic to develop emerging talent in AI research, engineering, and AI safety. It is designed to support individuals with strong technical potential, regardless of formal experience, and to provide them with the resources, mentorship, and funding required to conduct impactful empirical research in alignment with frontier AI safety priorities.

Anthropic is a public benefit corporation focused on building reliable, interpretable, and steerable AI systems that are safe and beneficial for society. The organization brings together researchers, engineers, policy experts, and operational leaders to advance the long-term development of safe artificial intelligence systems.

Program Structure and Objectives

The Fellows Program is designed as an intensive, full-time research experience aimed at producing high-quality public outputs, such as research papers or technical reports. Fellows are expected to work on empirical projects using external infrastructure such as open-source models and public APIs.

Key objectives of the program include:

Advancing AI safety and security research through hands-on empirical work

Producing publishable research outputs (many past fellows have published papers)

Building talent pipelines into frontier AI research roles

Supporting interdisciplinary collaboration across AI-related fields

Encouraging exploration of scalable, real-world AI safety problems

In previous cohorts, more than 80% of fellows successfully produced research publications, demonstrating the program’s emphasis on execution and scientific contribution.

Program Duration, Compensation, and Logistics

The program runs for approximately four months on a full-time basis. Fellows are expected to work around 40 hours per week, with the possibility of extension based on performance and project needs.

Participants receive structured support, including:

Weekly stipend:

3,850 USD

2,310 GBP

4,300 CAD

Research funding for compute and infrastructure (~$15,000/month)

Access to shared workspace locations in:

Berkeley, California

London, United Kingdom

Optional remote participation for eligible candidates in the US, UK, or Canada

Access to mentorship from leading researchers at Anthropic

Integration into a broader AI safety and research community

Visa sponsorship is not provided for fellows, and participants must already have work authorization in the US, UK, or Canada.

Application Timeline and Process

Applications are reviewed on a rolling basis, with structured cohorts beginning periodically. The next cohort begins on July 20, 2026, and applications must be submitted by April 26, 2026 for consideration.

The interview and selection process typically includes:

Initial application screening

Reference checks

Technical assessments and interviews

Research-focused discussions with potential mentors

Applicants are encouraged to apply even if they do not meet every listed qualification, as the program values potential, motivation, and research curiosity over rigid credential requirements.

Fellowship Workstreams

The program is organized into multiple specialized workstreams. Applicants may be considered across all areas based on their skills and preferences.

AI Safety Fellows

This workstream focuses on reducing catastrophic risks from advanced AI systems.

Research areas include:

Scalable oversight of advanced models

Adversarial robustness and AI control

Mechanistic interpretability of model internals

Model organisms of misalignment

AI welfare and evaluation frameworks

Typical profiles include candidates with:

Experience in empirical machine learning research

Strong Python programming ability

Interest in interpretability or alignment problems

Contributions to open-source AI research

AI Security Fellows

This track focuses on identifying and mitigating security vulnerabilities in AI systems.

Key focus areas:

Offensive security and vulnerability research

Red teaming and adversarial testing

LLM security and safety evaluation

Bug bounty experience and CVE reporting

Ideal candidates often demonstrate:

Experience in cybersecurity or penetration testing

Strong open-source contributions in ML or security domains

Ability to solve ambiguous technical problems independently

ML Systems & Performance Fellows

This stream emphasizes infrastructure, scalability, and systems-level ML engineering.

Work may include:

Building high-performance ML systems

Developing simulation environments for AI workloads

Optimizing training and inference pipelines

Supporting infrastructure-heavy research projects

Strong candidates typically have:

Experience with distributed systems and ML infrastructure

Engineering expertise in large-scale computing systems

Ability to balance research and production-grade engineering

Reinforcement Learning Fellows

This track focuses on reinforcement learning research and applied experimentation.

Key areas include:

RL environments for model training

Generalization studies in reinforcement learning

Model-based tools for training data analysis

Algorithm development and experimentation

Ideal profiles include:

Strong ML systems engineering experience

Familiarity with training and fine-tuning models

Ability to debug complex model training processes

Economics & Societal Impacts Fellows

This workstream explores the broader societal implications of AI systems.

Research topics include:

Economic impacts of AI adoption

Labor market and workforce transformation studies

Human-AI collaboration analysis

Model evaluation for societal well-being

Policy-relevant empirical research

Strong candidates often have:

Background in economics, social science, or data analysis

Strong writing and communication skills

Ability to interpret ambiguous empirical results

Interest in AI policy and societal impact

Candidate Profile and Requirements

Across all workstreams, ideal candidates typically demonstrate:

Strong motivation to improve AI safety and societal outcomes

Technical proficiency in Python programming

Ability to work full-time during the program duration

Experience in computer science, mathematics, physics, or related disciplines

Comfort working in fast-paced collaborative research environments

Strong communication and execution skills

Additional advantages include:

Open-source contributions

Prior ML or systems research experience

Domain expertise relevant to a chosen workstream

Program Philosophy and Values

Anthropic’s research approach is grounded in the belief that AI safety is a critical global challenge requiring large-scale, collaborative scientific effort. The organization treats AI research as an empirical science, similar in rigor and methodology to physics or biology.

Core values include:

Focus on long-term AI safety and alignment

Emphasis on interpretability and steerability of AI systems

Collaborative, cross-disciplinary research culture

Commitment to transparency and public research dissemination

Inclusion of diverse perspectives in shaping AI futures

For more opportunities such as these please follow us on Facebook, Instagram, WhatsApp, Twitter, LinkedIn and WPChannel

Disclaimer: Global South Opportunities (GSO) is not the organization offering the program. For any inquiries, please contact the official organization directly. Please do not send your applications & CVs to GSO, as we are unable to process them. Due to the high volume of emails, we receive daily, we may not be able to respond to all inquiries. Thank you for your understanding.

Related