Anthropic has opened applications for the Anthropic Residency, a paid, fixed-term research program for experienced technical leaders who want to move into full-time AI safety work. The application deadline is November 22, 2026.

The San Francisco-based company behind the Claude models is making an unusually broad recruiting pitch for a frontier AI lab: applicants do not need a machine learning background, a PhD, or a stack of published papers. Instead, the company is courting physics professors, technical founders, and staff-level engineers, offering prorated pay of $350,000 to $500,000 a year to those willing to pivot into AI safety research.

The pitch is aimed at technically accomplished professionals who have been looking to move into more meaningful work — a talent pool that extends well beyond the conventional machine learning research pipeline. In practice, the program treats AI safety research as a field that accomplished specialists can enter directly, rather than one accessible only through doctoral training and a publication record.

What the Residency Offers

Successful applicants join the lab for a fixed-term placement, with terms varying by track. Placements run anywhere from six weeks to six months, and the research summary describes three formal tracks: six weeks, three months, or six months.

Compensation is set at $350,000 to $500,000 on an annualized basis, prorated to match how long a resident stays — so someone on the six-week track earns a proportional slice of that band, not the full annual figure. Beyond pay, residents receive mentorship and working resources, including access to Claude models and GPU compute.

Location requirements depend on the duration of the placement. Shorter placements come with remote options, while longer ones may require residents to be based in San Francisco. Anthropic will sponsor visas for placements of three months or more, meaning international candidates are eligible for sponsorship only on the two longer tracks.

The research areas on the program's agenda include:

• AI safety and alignment
• Scalable oversight
• Interpretability
• AI security and red-teaming
• ML systems and reinforcement learning

Read together, the agenda spans work on understanding model internals (interpretability), methods for humans to supervise AI systems (scalable oversight), stress-testing systems for weaknesses (AI security and red-teaming), and the training techniques underneath them (ML systems and reinforcement learning).

Who Anthropic Is Looking For

The target list is specific. Anthropic named professors in quantitative disciplines, technical founders, and staff-level engineers as the kinds of people it wants to hear from. The research summary points to fields such as physics, mathematics, and neuroscience as examples of the backgrounds the company is courting.

Once applications close on November 22, the timeline is fairly quick. Offers are expected to go out in January 2027 — a concrete decision point on the calendar for professionals weighing a mid-career pivot.

Why Anthropic Is Doing This Now

The program arrives during a period of fast growth for the company. Anthropic reportedly brought in around $17 billion in revenue for the third quarter of 2026, according to the research findings, with net losses narrowing. The company is also preparing for a planned IPO.

The Residency sits alongside the fellowships and academies Anthropic runs to build up its AI safety and research bench, adding another channel through which the lab can bring senior technical talent into its safety work.

Disclosure: This article is based on a CryptoBriefing report edited by Diego Almada Lopez.