Anthropic Residency 2026: Paid AI Safety Research Program for Established Technical Leaders – Applications Open

The Anthropic Residency 2026 is now accepting applications from experienced technical professionals, researchers, professors, founders, and engineers who want to transition into full-time artificial intelligence (AI) research. This paid, fixed-term research program provides an opportunity to work on critical AI safety, alignment, interpretability, machine learning systems, and frontier AI research problems alongside leading researchers at Anthropic. The Anthropic Residency is specifically designed for individuals who have already demonstrated leadership in a technical field and are interested in applying their expertise to the development of safe and beneficial artificial intelligence. Unlike many traditional AI research programs, applicants do not necessarily need previous machine learning experience, a PhD, or published academic research papers to be considered. The program offers research placements lasting six weeks, three months, or six months, depending on the participant's preferences and the program's requirements. Residents receive mentorship from experienced Anthropic researchers, access to advanced AI models and computing resources, dedicated time to learn modern machine learning, and compensation equivalent to $350,000–$500,000 per year. Applications for the current round of the Anthropic Residency close on November 22, 2026. This opportunity is particularly relevant for established technical professionals who want to move into AI research, contribute to the safety of increasingly capable AI systems, and build a long-term career in artificial intelligence research. Overview of the Anthropic Residency 2026 The Anthropic Residency is a paid research program intended to help experienced technical professionals transition into full-time AI research. The program focuses on empirical research questions and provides opportunities to collaborate with Anthropic's research teams. Residents are matched with research problems based on their interests, professional backgrounds, and technical expertise. The following table summarizes the major details of the program. Opportunity Detail Information Program Name Anthropic Residency Organization Anthropic Program Type Paid, fixed-term research program Research Focus AI safety, AI alignment, interpretability, AI security, machine learning systems, and reinforcement learning Duration Six weeks, three months, or six months Compensation $350,000–$500,000 per year equivalent Preferred Location Anthropic offices in San Francisco, California, USA Remote Participation Possible, particularly for shorter placements Application Deadline November 22, 2026 Academic Requirement PhD not required Previous Machine Learning Experience Not required Published Research Papers Not required Application Method Online application through the US or UK job posting What Is the Anthropic Residency? The Anthropic Residency is a research initiative designed to bring experienced technical professionals into the field of AI safety and empirical AI research. Artificial intelligence is developing rapidly, and the growing capabilities of advanced AI systems create important technical challenges. Researchers need to understand how these systems behave, how they can be made more reliable, how their internal mechanisms operate, and how potential risks can be identified and mitigated. Anthropic established this Residency to help experienced professionals from other technical disciplines develop the knowledge and skills required to contribute to these challenges. The program recognizes that valuable AI research skills are not limited to individuals who have spent their entire careers studying machine learning. For example, a mathematician may have strong analytical skills, a physicist may have experience developing complex models, a software engineer may understand large-scale technical systems, and a neuroscientist may have experience studying complex behavior. These capabilities can be valuable in AI research. The Residency provides an opportunity for professionals with these backgrounds to learn modern machine learning and apply their existing technical expertise to research problems involving advanced AI systems. Why Anthropic Is Recruiting Experienced Technical Leaders Machine learning is a relatively young research field. Many established researchers in AI originally studied or worked in other technical disciplines before moving into machine learning. Anthropic emphasizes that the ability to frame important research questions, develop a research agenda, evaluate ideas, and determine which problems deserve attention is increasingly important in advanced AI research. Experienced technical leaders may already possess these capabilities. The Residency is designed to help such professionals build the additional understanding of modern machine learning and AI safety needed to contribute effectively. The program builds on the experience of the Anthropic Fellows Program, which demonstrated that researchers entering the field from different backgrounds can make meaningful contributions to AI research. Previous Fellows have worked on research involving subliminal learning, the Petri auditing tool, and persona vectors. The new Residency adapts this approach for established technical leaders who have already demonstrated substantial technical achievement. Benefits of the Anthropic Residency 2026 The Anthropic Residency offers several benefits for professionals seeking to transition into AI research. 1. Work on Critical AI Research Problems Residents will work on research problems developed by Anthropic researchers and matched to their interests and professional backgrounds. The program focuses on empirical research, allowing participants to investigate the behavior, capabilities, limitations, and safety properties of advanced AI systems. Research opportunities may involve AI alignment, scalable oversight, adversarial robustness, interpretability, AI security, machine learning infrastructure, and reinforcement learning. This allows residents to contribute to technical questions that are important for the development of safe and beneficial AI. 2. Mentorship from Leading AI Researchers Each resident will be paired with an Anthropic researcher who will provide regular one-on-one mentorship. This mentorship is intended to support participants as they transition into the field of AI research. Residents can receive guidance on research methodology, technical implementation, problem selection, and the development of research projects. Potential mentors include researchers working across Anthropic's Alignment Training, Alignment Science, Alignment Stress-Testing, Scalable Oversight, Interpretability, Model Psychology, and AI Security teams. The mentorship structure is particularly valuable for professionals who have strong technical expertise but limited previous experience in machine learning research. 3. Dedicated Support for Learning Machine Learning Applicants are not expected to arrive as machine learning experts. The program includes dedicated self-study time, research talks, and guidance to help residents develop an understanding of modern machine learning and AI safety research. This learning support is designed to help participants build the knowledge required to work effectively on empirical AI research projects. For professionals transitioning from mathematics, physics, neuroscience, software engineering, or other quantitative disciplines, this can provide an important foundation for entering AI research. 4. Access to Advanced AI Models and Computing Resources Residents will have access to frontier AI models and a dedicated computing budget for their research projects. Access to computing resources is important because empirical AI research often requires experiments involving advanced models, training systems, evaluation pipelines, and large-scale computational infrastructure. The dedicated compute budget allows residents to carry out research projects that require substantial technical resources. 5. Flexible Research Duration The Anthropic Residency offers three possible placement durations: • Six weeks • Three months • Six months Applicants will be asked about their preferred duration during the application process. Anthropic will also aim to arrange flexible start dates. This flexibility may be useful for established technical professionals who want to transition into AI research while considering their existing professional commitments. However, applicants should understand that the actual placement duration and start date will depend on the program's arrangements. 6. High-Level Compensation The Residency provides compensation equivalent to $350,000–$500,000 per year. This compensation reflects the seniority and experience of the professionals Anthropic is seeking. The program is intended for individuals who have already led ambitious technical work and possess substantial expertise in their respective fields. The stated compensation is an annual equivalent. It should not be interpreted as a guaranteed payment of $350,000–$500,000 for every Residency placement. The actual compensation for a fixed-term placement will depend on its duration and the applicable arrangements. 7. Opportunity to Pursue a Full-Time AI Research Career The Residency is designed to help participants transition into full-time AI research. By the end of the program, Anthropic aims to position every resident to continue working on AI research full-time, either at Anthropic or elsewhere. For professionals seeking to move into AI safety and alignment research, this can provide a structured pathway to develop relevant expertise, conduct empirical research, and build connections with established researchers. The program does not guarantee a permanent position at Anthropic. However, it is intended to help participants become better prepared for long-term research careers. Research Areas Available Through the Anthropic Residency Residents will collaborate with Anthropic's research teams on empirical research questions. The program is particularly interested in several areas of AI research. 1. AI Safety and Alignment AI safety and alignment research focuses on developing AI systems that behave in accordance with intended objectives, remain reliable, and avoid harmful outcomes. As AI systems become increasingly capable, researchers need methods to ensure that their behavior remains safe and beneficial. The Anthropic Residency includes several research directions within this field. Scalable Oversight Scalable oversight focuses on developing techniques that allow increasingly capable AI systems to remain helpful and honest, even when their capabilities exceed human-level performance in particular domains. As AI systems become more advanced, directly supervising every action or output may become increasingly difficult. Research in scalable oversight investigates methods that can improve the effectiveness of human supervision and support reliable AI behavior. Adversarial Robustness and AI Control This research area focuses on ensuring that advanced AI systems remain safe and harmless in unfamiliar, adversarial, or unexpected situations. Researchers investigate how AI systems respond to difficult inputs, adversarial conditions, and situations that may expose weaknesses in their behavior. The objective is to develop methods that improve system robustness and maintain appropriate control over advanced AI systems. Model Organisms of Misalignment Model organisms of misalignment involve developing experimental systems that help researchers understand how alignment failures may arise. These systems can be used to investigate potential causes of misalignment and study how undesirable behaviors develop. The research aims to improve the empirical understanding of AI alignment failures and support the development of better mitigation t

• Established technical background or leadership in a technical field • Must be an experienced technical professional, researcher, professor, founder, or engineer • PhD not required • Previous machine learning experience not required • Published research papers not required