Anthropic opens residency for technical leaders to join AI safety research

A bridge to AI safety research

Anthropic has officially opened applications for its 2026 Residency program, a structured initiative designed to bring experienced technical leaders into the field of artificial intelligence. The program aims to help established professionals transition into full-time research roles specifically focused on creating safer and more beneficial AI systems. By offering fixed-term placements, the company wants to provide a clear pathway for experts to apply their existing skills to frontier research challenges.

The residency is built on the belief that AI research benefits from a wide variety of scientific and technical disciplines. Anthropic notes that many of the most successful AI researchers originally built their expertise in fields like physics, mathematics, neuroscience, and software engineering. This program seeks to replicate that diversity by recruiting leaders who can identify important scientific questions and develop effective research strategies.

As reported by Global South Opportunities, the residency provides a variety of flexible placement options to suit different professional needs. Participants can choose to join the program for six weeks, three months, or six months depending on their availability. This flexibility allows high-level professionals to engage with the company's research goals without necessarily committing to a permanent role immediately.

Mentorship and technical resources

Selected residents will be paired with experienced Anthropic research mentors to provide individual guidance throughout their placement. These mentors include leaders of major initiatives such as Alignment Training, Alignment Science, and Scalable Oversight. This ensures that participants have direct access to the people currently defining the standards for how AI models should behave and be governed.

Beyond human mentorship, the program provides residents with the high-level tools required to conduct serious empirical research. This includes dedicated computing budgets and access to frontier AI models, which are often prohibitively expensive for independent researchers. Residents also receive dedicated self-study time and access to research talks to help them quickly gain proficiency in modern machine learning techniques.

The program is designed to accelerate the transition of the participants into full-time research careers. While the residency is a fixed-term commitment, the goal is to prepare these individuals for long-term roles in the field. These roles could eventually be held at Anthropic or at other organizations, helping to build a broader community of experts capable of handling AI safety risks.

Targeting diverse technical leaders

Anthropic is specifically looking for individuals who have demonstrated the ability to lead ambitious technical projects. The ideal candidates include technical founders with significant development experience, professors in quantitative disciplines, and staff-level software engineers. The company is interested in people who can solve complex problems, communicate clearly, and implement ideas rapidly.

One of the most notable aspects of the residency is that previous machine learning experience or a PhD is not required for application. This lowers the barrier for experts in other fields who possess the underlying logic and engineering rigor needed for AI work. Anthropic is also actively encouraging applications from underrepresented groups in the technology sector to ensure a variety of perspectives in their research teams.

The program seeks to tap into the unique strengths of different professional backgrounds to tackle specific research priorities. For example, those with a background in neuroscience might contribute to model psychology or interpretability. Meanwhile, software engineers might focus on building AI auditing tools or improving the underlying hardware infrastructure and optimization methods.

Core research priorities

Residents will be matched with research opportunities that align with their specific interests and professional expertise. Key areas of focus include scalable oversight, which involves developing methods to keep advanced AI systems helpful and honest. Other priorities include adversarial robustness, which tests how AI systems perform in challenging or hostile environments.

The program also explores more specialized topics such as model organisms and AI welfare. Model organisms involve studying controlled examples of misalignment to understand how AI failures might occur in the future. Additionally, researchers will investigate AI welfare considerations to develop appropriate evaluations for the well-being of potential AI systems.

Security and red-teaming also play a major role in the residency's research goals. Participants may work on identifying security risks across various domains, including cybersecurity, robotics, and cryptography. By focusing on these diverse areas, Anthropic aims to ensure that the development of advanced AI remains safe, reliable, and beneficial for society.

Application details and timeline

The application window for the 2026 residency is currently open, with a closing date set for November 22, 2026. Interested candidates can apply through regional recruitment postings on the official Anthropic careers page for the United States and the United Kingdom. The company aims to accommodate different start dates to make the program accessible to a wider range of professionals.

Compensation for the residency is highly competitive, with residents receiving pay equivalent to $350,000–$500,000 annually. While the preferred working location is Anthropic’s offices in San Francisco, the company may allow for remote participation for some of the shorter placement options. This high level of compensation reflects the seniority of the roles and the importance of the research being conducted.

Candidates who apply for the residency may also be considered for the Anthropic Fellows Program. This separate initiative has previously enabled emerging researchers to contribute to projects involving subliminal learning and AI auditing. By opening these doors, Anthropic is creating a multi-tiered system to bring both established leaders and emerging talent into the AI safety space.