Anthropic Residency 2026: $350,000-$500,000 AI Safety Research Programme, Apply by 22 November

Anthropic opened applications on 2 November 2026 for the Anthropic Residency, a paid, fixed-term research programme for established technical leaders who want to move into full-time work on AI safety. The company said applications for this round close on 22 November 2026, and the programme offers placements of six weeks, three months, or six months, paid at $350,000 to $500,000 a year on a pro-rated basis.

Read the eligibility bar before anything else, because this is not an entry-level scheme and not a general graduate intake. Anthropic said it wants people who have already led ambitious work in a technical field: professors in quantitative disciplines, technical founders, or staff-level engineers. A background in machine learning is not required, and the company said applicants do not need a PhD, prior machine learning experience, or published papers. The United States posting lists a minimum of a bachelor's degree or an equivalent combination of education, training, or experience, and says the years of experience required will match the internal level of the role.

At a glance	Details
Programme	Anthropic Residency
Organiser	Anthropic
Placements	Six weeks, three months, or six months; other lengths between three and six months possible by exception
Pay	$350,000 to $500,000 a year equivalent, pro-rated for the placement, plus benefits
Application deadline	22 November 2026, 11:59 PM Pacific Time
Offers expected	January 2027
Location	San Francisco preferred; remote participation possible, especially for shorter placements
Visa sponsorship	Sponsored for placements of three months or longer; six-week placements are encouraged to be remote from a country where Anthropic can hire
How to apply	Through the United States or United Kingdom posting linked from the official announcement

Who the Anthropic Residency is for

Anthropic framed the round as a way into AI safety for senior people from neighbouring fields. The company said many of its senior researchers started out as physicists, mathematicians, neuroscientists, or software engineers, and that framing a research agenda and judging what is worth doing matter more as model capabilities improve. The Residency is built for people who already show those strengths in another technical field and need structured time, mentorship, and compute to apply them to frontier AI.

The announcement lists the qualities it is looking for in plain terms: a record of leading ambitious technical work, motivation to make AI safe and beneficial for society, enthusiasm for moving into empirical AI research full time, comfort in fast-paced collaborative work, and the ability to put ideas into practice quickly and explain them clearly. The United States posting lists research papers and large language model experience as strengths, not fixed requirements, and the company said applicants should not rule themselves out for missing every single qualification.

What residents will work on

Residents work on problems set out by Anthropic researchers and matched to their interests and background, with a concrete, scoped project and direct mentorship. Each resident is paired with an Anthropic mentor and receives regular one-on-one guidance, dedicated self-study time, talks, and support to get up to speed on modern machine learning and AI safety research. Residents also receive a dedicated compute budget and access to frontier models.

The research areas named in the announcement fall into three groups. The first is AI safety and alignment, covering scalable oversight, adversarial robustness and AI control, model organisms of misalignment, model internals and interpretability, safeguards research, and AI welfare. The second is AI security and frontier red-teaming, meaning stress-testing frontier models to map their capabilities in fields such as cybersecurity, robotics, and cryptography. The third is machine learning systems and reinforcement learning, including model architecture, optimisers, training data, accelerator simulators, hardware backends, and reinforcement learning environment pipelines. Mentors named for the round lead teams in alignment training, alignment science, stress-testing, scalable oversight, interpretability, model psychology, adversarial machine learning, and safety and pretraining.

Pay, place and visas for the Anthropic Residency

Pay is set at $350,000 to $500,000 a year equivalent, pro-rated for the length of the placement, plus benefits. A three-month placement therefore works out at roughly one quarter of the annual figure before tax. The programme is full time for the duration of the placement, but that is an aim, not a promise of a permanent post.

Place and visas need a careful read for applicants outside the United States. Anthropic said it prefers residents to work from its San Francisco offices, where most of its research staff are based, but remote participation is possible, especially for shorter placements. Remote participants need to be based in a country where Anthropic can hire. Visa sponsorship is offered for placements of three months or longer, while six-week placements are encouraged to be remote from a country where Anthropic can hire. The announcement does not state how many residents this round will take, so applicants should not assume a fixed intake number.

How to apply for the Anthropic Residency

Applications go through the United States or United Kingdom posting linked from the official announcement, and applicants should use the posting that matches their location. The United States posting sets out the interview process as an initial application, a recruiter screen, a technical assessment, a research discussion, and a reference check. Applications close on 22 November 2026 at 11:59 PM Pacific Time, and the company said it expects to make offers in January 2027.

Applicants should apply only through the official announcement or the linked postings, and should treat any other route with care. Anthropic said its recruiters only contact applicants from anthropic.com email addresses, and that legitimate recruiters will never ask for money, fees, or banking information before the first day of work. Any message asking for payment to secure a Residency place should be treated as a fraud attempt.

What to check before applying

Three points deserve a direct check before an applicant commits. First, the programme is fixed term, so applicants who need a permanent role should weigh that against the senior pay and the research access on offer. Second, the San Francisco preference matters: applicants who cannot relocate should confirm whether their preferred placement length can genuinely be done remotely from their country. Third, no start date is stated beyond flexible timing, so applicants with notice commitments should raise timing at the recruiter screen.

For readers tracking paid routes into AI work, AI startup jobs across 1,100 companies cover permanent roles at a median base of $190,000 a year, while AI training jobs for consultants pay by the hour. Readers may also want Periscope Labs Senior Applied AI Engineer post in Dhaka.

Sources: Anthropic Alignment Science Blog, "Anthropic Residency: applications open for established technical leaders", published 2 November 2026, at https://alignment.anthropic.com/2026/anthropic-residency/; Anthropic United States posting, "Anthropic Research Resident", at https://job-boards.greenhouse.io/anthropic/jobs/5438311008. The announcement states the programme, pay range, placement lengths, deadline, and location preference quoted here; the number of residents this round will take is not stated.