Pivotal AI Safety Research Fellowship 2027: 15 Weeks in London with a Stipend of 15000 to 18000 Pounds

Do you want to work on making advanced AI safer? The Pivotal AI Safety Research Fellowship gives you 15 weeks of full time research in London. You work with an expert mentor and a research manager. You also get a stipend of 15000 to 18000 pounds.

This round is bigger than before. Pivotal has stretched the programme from 9 weeks to 15 weeks and raised its housing support and stipends. Applications close on 1 November 2026.

At a Glance

Last Information checked: 28 October 2026

What the Fellowship Is

This is a research programme for people who want to work on the safety and governance of advanced artificial intelligence. AI systems are getting more capable. That raises hard questions about reliability security governance and alignment. Pivotal supports people who want to study these questions and find practical answers.

For 15 weeks you work in person on a research project with an external mentor. A research manager helps you set clear goals and stay on track. By the end Pivotal wants you to have a published research output and clear next steps for your career.

Also Read: Asian Cultural Council New York Fellowship 2028: Six Months in New York (Fully Funded)

Fellowship Benefits

• Stipend of 15000 to 18000 pounds for the 15 weeks.
• Travel support
• Housing support of 1800 pounds a month for fellows based outside London
• Research budget. Pivotal provides budget for food, housing, travel and the compute your research needs
• Mentor. Weekly one to one meetings with an experienced researcher who gives guidance and feedback
• Research manager. Help to set realistic goals solve problems and plan your next career steps
• Workspace. An in person research space at the London Initiative for Safe AI with lunch and dinner on weekdays

Research Areas

Fellows work in three broad areas.

• Technical AI safety.
• Technical AI governance.
• AI policy and governance.

Past projects include AI agent compliance with sanctions law and evaluation of models that may be deceptive. They also include disclosure frameworks for AI lab insiders and privacy protecting monitoring of AI computing infrastructure. Use them as examples of the range and not as a list to copy.

Who Should Apply

The fellowship may suit you if you are:

• An early career researcher interested in AI and its effect on society
• A computer science graduate interested in AI security, interpretability or model evaluation
• Someone with a background in mathematics, statistics or machine learning
• A researcher interested in technology policy and AI governance
• A professional who wants to move into AI safety research or governance

Mentors

Pivotal’s mentors work across AI research technical development and policy. Browse the mentor list on the Pivotal website before you apply. Pick mentors whose work matches your project idea. Your answers should show why your skills fit their research.

Key Dates

• 1 November 2026 at 23:59 Anywhere on Earth time: Application deadline
• 18 January 2027: Fellowship starts in London
• 30 April 2027: Fellowship ends

How To Apply

1. Go to the Pivotal Research website
2. Open the AI Safety Fellowship section.
3. Browse the mentor list and choose the mentors you would like to work with.
4. Fill in the main application form.
5. Complete a short mentor specific section for each mentor you choose. Pivotal says each one takes about 20 minutes.
6. Submit before 1 November 2026 at 23:59 Anywhere on Earth time.

Visit Here to Apply

How Selection Works

Shortlisted applicants may go through three stages.

• Video interview. A short recorded interview of under 10 minutes
• Research task. A mentor specific exercise that takes about one to three hours
• Personal interview. A talk with the team or possible mentors that lasts about 30 to 45 minutes

Check the official instructions for the exact current steps.