If you are into AI research and you keep hearing the name "Anthropic" everywhere, you are not alone. Anthropic, the company behind the Claude AI models, has built one of the most talked-about research fellowships in the world right now. It is called the Anthropic Fellows Program, and it is made for people who want to spend real time working on AI safety without needing a PhD or years of published papers to prove themselves. Think of it as a paid, full-time research internship, but at a much higher level. It runs for four months, and during that time, fellows work closely with experienced Anthropic researchers on real, unsolved problems in AI safety. The goal is simple: by the end of four months, you should have produced something public, usually a research paper or a technical report that adds real value to the field. This is not a classroom course. There are no lectures. You are placed directly into the research process, working on a project that an Anthropic mentor has pitched, and then you and your mentor shape that project together based on your interests and skills.

Key Details: Anthropic Fellows Program,

• Program: Anthropic Fellows Program

• Duration: 4 months, full-time (possible extension)

• Stipend: ~3,850 USD per week (around 65,000 USD total)

• Compute/Research Budget: ~15,000 USD per month

• Location: Berkeley, California or London, UK (shared workspace); remote option for US, UK, and Canada

• Eligibility: No PhD or prior ML experience required; strong Python skills and motivation matter most

• Application: Rolling basis, multiple cohorts per year

British Academy Postdoctoral Fellowship 2026 in UK

How Much Does It Pay?

Fellows get a weekly stipend of around 3,850 USD, which adds up to roughly 65,000 USD across the full four months. On top of that, there is a separate monthly budget of about 15,000 USD just for compute power and other research expenses, so you are not paying out of pocket to run experiments.

Fellows also get access to a shared workspace, either in Berkeley, California, or in London, UK. Remote participation is open to people based in the US, UK, and Canada.

Research Tracks, Where Fellows Work

Originally, this program focused only on AI safety. But Anthropic has since expanded it into several specialized tracks because the company realized safety research touches many different fields. Right now, the work streams include the following:

• AI Safety Fellows – the original and largest track, focused on reducing catastrophic risks from advanced AI systems. This includes work on scalable oversight, adversarial robustness, interpretability, and model behavior.

• AI Security Fellows – focused on how AI can be misused for cyberattacks and how to defend against new types of jailbreaks and exploits.

• ML Systems and Performance Fellows – for people who enjoy building large-scale infrastructure and optimizing how massive models actually run.

• Reinforcement Learning Fellows – working on training environments, RL algorithms, and evaluation methods.

• Economics and Societal Impacts Fellows – for those more interested in the policy, economic, and social side of AI rather than pure engineering.

Who Can Apply? What is the Eligibility?

Anthropic cares far more about your ability to actually do research than about your credentials. You do not need a PhD. You do not need published papers. You do not even need prior machine learning experience.

What they do look for is the following:

• Solid coding skills, especially in Python

• The ability to take a vague, messy problem and make real progress on it

• Genuine motivation to reduce risks from advanced AI systems

• A track record of learning fast and shipping work, even under uncertainty

People from physics, mathematics, cybersecurity, and other quantitative fields have all succeeded in this program, so do not count yourself out just because your background is not pure computer science.

MIF Research Fellowship: Fully Funded

What Have Past Fellows Actually Done?

1. Past fellows have not just written theoretical papers; they have produced work with real, measurable impact. One project led to fellows building agents that uncovered more than 4.6 million dollars worth of vulnerabilities in blockchain smart contracts.

2. Another group developed a method that helped block entire categories of dangerous jailbreak attacks before they could spread. Others worked on open-source tools that let researchers trace how large language models "think" internally, step by step.

3. In the very first cohort, more than 80 percent of fellows ended up producing a public research output, and over 40 percent were later hired full-time by Anthropic. That is an incredibly strong outcome for any research fellowship.

How to Apply: Step by Step

Applications are reviewed on a rolling basis, and new cohorts start a few times a year, so there is no need to wait for a single fixed deadline to get started.

Step 1: Pick your workstream. Before you apply, take a few minutes to decide which track fits you best: AI Safety, AI Security, ML Systems and Performance, Reinforcement Learning, or Economics and Societal Impacts.

Step 2: Submit the written application. This is where you share your background, your motivation, and why you want to work on AI safety. Keep it honest and specific.

Step 3: Wait for the initial review. Applications are read on a rolling basis. If your profile looks like a good fit, you will be moved forward to the next stage.

Step 4: Complete the technical assessment. If you advance, you will usually be given a coding or research-style task to complete. This step exists to see how you actually approach a problem, not to test memorized knowledge.

Step 5: Interview rounds. Candidates who pass the technical stage move on to interviews with Anthropic researchers. These conversations dig into your technical thinking, your motivation, and how you might approach an open-ended research problem.

Step 6: Project and mentor matching. If you are offered a spot, you will go through a matching process where Anthropic mentors pitch project ideas, and you choose and shape one together based on your interests.

Step 7: Start the fellowship. Once matched, you begin your four months of full-time research, with weekly stipend payments, your compute budget, and direct mentorship from day one.

Apply Now: The $300,000 LEGO Foundation Global Fellowship 2027 is Officially Open

Frequently Asked Questions

Do I need a PhD to apply? No. Anthropic has said clearly that they care more about your ability to do research than your formal credentials. Many past fellows came in without a PhD.

Do I need prior machine learning experience? Not necessarily. Strong coding skills, especially in Python, and the ability to make progress on ambiguous problems matter more than a specific ML background. Fellows have come from physics, mathematics, and cybersecurity, among other fields.

Is the fellowship remote or in-person? Both are possible. Anthropic provides shared workspaces in Berkeley, California, and London, UK, and also supports remote fellows based in the US, UK, and Canada.

How long does the fellowship last? Four months, full-time, with the possibility of extension in some cases.

Can I apply if I am based outside the US, UK, or Canada? Remote participation is currently limited to the US, UK, and Canada, so check the latest application page for your eligibility before applying, since these details can change between cohorts.

What happens after the fellowship ends? Many fellows go on to publish their research publicly. In the program's first cohort, over 40 percent of fellows were later hired full-time by Anthropic, and others have gone on to safety roles at other organizations.

Is there an age limit or specific degree requirement? No specific age limit is mentioned. What matters most is your technical fundamentals and your motivation for the work, not a particular degree or background.

How competitive is the program? It is highly competitive, given the strong pay, mentorship, and the chance to work on cutting-edge AI safety research. A clear, specific application and any demonstrated project work can help you stand out.

Always apply through Anthropic's official careers page rather than third-party sites, since application links and deadlines can change between cohorts. A link is provided below.