GizPulse

Anthropic, with Constellation

Anthropic Fellows Program January 2027: Paid AI Safety Research With a $3,850 Weekly Stipend

ScholarshipsBerkeley, USA; London, UK; or remote in the US, UK or CanadaDeadline October 18, 20263 min read0 comments
Inside Anthropic's San Francisco office during a visit by UK Science Secretary Peter Kyle in October 2024

Photo by Alecsandra Dragoi / Department for Science, Innovation and Technology, CC BY 2.0, via Wikimedia Commons

Opportunity Overview

Anthropic's four-month Fellows Program pays $3,850 a week plus about $15,000 a month in compute for AI safety and security research. You need work authorization in the US, UK or Canada. Apply by 18 October 2026.

Apply Now

Full Program Details

Most people who want to work on AI safety hit the same wall: the best research teams want experience, and you cannot get experience without joining a research team. The Anthropic Fellows Program is one of the few paid routes around that wall. It gives promising technical people four months, a mentor from Anthropic, serious compute and a weekly stipend to produce real research, regardless of their previous experience.

Applications for the January 2027 cohort are open now and close on 18 October 2026. Anthropic says over 80% of fellows in one earlier cohort produced papers, and 25% to 50% of fellows in previous cohorts received a full-time offer.

Fellows receive a weekly stipend of 3,850 USD, 2,310 GBP or 4,300 CAD, plus around $15,000 a month in compute funding and direct mentorship from Anthropic researchers.

Details at a glance:

  • Host: Anthropic, with Constellation as recruiting and programme partner
  • Cohort: January 2027
  • Length: 4 months, full-time (about 40 hours a week)
  • Stipend: 3,850 USD, 2,310 GBP or 4,300 CAD per week, plus benefits that vary by country
  • Compute: about $15,000 a month, plus other research expenses
  • Workspaces: Berkeley, California, or London, with remote options in the US, UK and Canada
  • Deadline: 18 October 2026

What the Fellowship Offers

  • Four months of full-time, funded research on an empirical project tied to Anthropic's research priorities
  • Direct mentorship from Anthropic researchers
  • A shared workspace in Berkeley or London
  • Funding for compute and other research expenses
  • A public output at the end, such as a paper submission
  • Connections to the wider AI safety and security research community

There are two workstreams. AI Safety Fellows work on areas such as scalable oversight, adversarial robustness and AI control, model organisms of misalignment, mechanistic interpretability and AI welfare. AI Security Fellows work on security research, with listed mentors including Nicholas Carlini.

Who Can Apply

  • People with a strong technical background in computer science, mathematics or physics
  • Fluent Python programmers
  • People available to work full-time for the four months
  • Applicants who already have, or can independently obtain, full-time work authorization in the US, UK or Canada, and who will be located in that country during the programme

Anthropic does not sponsor visas for fellows. That is the single biggest filter, so check it before you start. If you live in Nigeria, Ghana or Kenya without work rights in one of those three countries, this round is not open to you.

Previous research experience is not required. Anthropic says it wants people who are excited to move into empirical AI research and encourages candidates to apply even if they do not meet every listed qualification.

How to Apply

  1. Read the official listing and decide between the AI Safety and AI Security workstreams.
  2. Confirm you have work authorization in the US, UK or Canada for the full programme.
  3. Submit the application through the official Greenhouse page before 18 October 2026. Constellation manages applications and interviews.
  4. Complete the reference check and technical assessments if you move forward.
  5. Take part in technical interviews and a research discussion.
  6. If selected, go through project selection and mentor matching before the cohort starts.

GizPulse Verdict

Who this is for: Strong coders and quantitative researchers in the US, UK or Canada who want a paid, mentored entry into AI safety or AI security research, including career changers from software engineering, security or physics.

Who should pause: Anyone without existing work rights in the US, UK or Canada, since there is no visa sponsorship. Also pause if you cannot commit four full-time months.

One tip to stand out: Show, do not tell. Link a public repo or write-up where you ran an experiment on a language model, measured something and explained what surprised you. Anthropic lists open-source contributions and hands-on LLM work among the strongest signals.

Read the full listing and apply on Anthropic's official Fellows Program page.

Explore More On These Topics

FellowshipBerkeley, USA; London, UK; or remote in the US, UK or Canada

Share This Story

Get GizPulse Weekly

Receive jobs, opportunities, and practical tech insights every Sunday.

Please complete verification to subscribe.

Comments

Comments are moderated and published after approval.

Please complete verification before posting your comment.

No comments yet.

Related Opportunities