Staff+ Researcher, Cybersecurity Products

About the Role:

We're looking for a Capabilities Researcher to join the team building Claude Security. In this role, you'll identify which security capabilities in frontier models are ready to build on, measure how well they perform, and work out how to make them useful to customers who are not security experts.

Frontier models have become substantially more capable at security work over recent generations, and they continue to improve with each release. That progress creates a set of practical questions: which capabilities are reliable enough to depend on, how they perform in realistic conditions, how they behave when an adversary is involved, and where their limits are. You'll be responsible for answering those questions through fast prototyping and rigorous evaluation, and your findings will shape what the team decides to build.

You'll also work on making those capabilities usable. Together with engineers on the team, you'll design the scaffolding, tooling, and defaults that let a strong model capability do useful work for a non-expert, and you'll stay involved as it becomes a product.

This is a research role on a product team, with a broad charge and real latitude in what you investigate. It suits someone who already has ideas about what AI should be able to do for security teams and wants the models, the time, and the engineering support to pursue them.

Responsibilities:

• Prototype rapidly to find define the AI frontier for cybersecurity work
• Design evaluations that measure model performance on the work security teams actually do
• Build the datasets, harnesses, and scoring those evaluations depend on
• Engage with the cybersecurity community to help define where AI can make the most impact
• Work with engineers and researchers to operationalize promising capabilities into something customers can rely on
• Track how model capabilities for security are changing, and what that means for what we build next
• Share findings that inform product direction, and partner with product leadership on priorities

You may be a good fit if you:

• Have deep expertise in one or more security domains, such as vulnerability research, exploit development, reverse engineering, malware analysis, incident response, or offensive security
• Have built AI-powered tools or capabilities for security work
• Can get from an idea to a working prototype quickly, and abandon the ones that don't hold up
• Are comfortable designing rigorous evaluations and interpreting the results honestly
• Can write and communicate clearly about technical findings
• Have 7+ years of experience in security research, security engineering, or a closely related field

Strong candidates may also have:

• Published research, CTF results, CVEs, or open source security tooling
• Experience with model evaluation, benchmarking, or red teaming
• Experience building agentic applications
• Familiarity with the safety considerations of AI in security contexts

Compensation:

The annual compensation range for this role is $405,000 — $485,000 USD.

Logistics:

• Minimum education: Bachelor's degree or an equivalent combination of education, training, and/or experience
• Required field of study: A field relevant to the role as demonstrated through coursework, training, or professional experience
• Location-based hybrid policy: Currently, we expect all staff to be in one of our offices at least 25% of the time. However, some roles may require more time in our offices.
• Visa sponsorship: We do sponsor visas. However, we aren't able to successfully sponsor visas for every role and every candidate.

We encourage you to apply even if you do not believe you meet every single qualification. Not all strong candidates will meet every single qualification as listed.

How we're different:

We believe that the highest-impact AI research will be big science. At Anthropic we work as a single cohesive team on just a few large-scale research efforts. And we value impact — advancing our long-term goals of steerable, trustworthy AI — rather than work on smaller and more specific puzzles. We view AI research as an empirical science, which has as much in common with physics and biology as with traditional efforts in computer science. We're an extremely collaborative group, and we host frequent research discussions to ensure that we are pursuing the highest-impact work at any given time. As such, we greatly value communication skills.

The easiest way to understand our research directions is to read our recent research. This research continues many of the directions our team worked on prior to Anthropic, including: GPT-3, Circuit-Based Interpretability, Multimodal Neurons, Scaling Laws, AI & Compute, Concrete Problems in AI Safety, and Learning from Human Preferences.

Come work with us!

Anthropic is a public benefit corporation headquartered in San Francisco. We offer competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space in which to collaborate with colleagues. Guidance on Candidates' AI Usage: Learn about our policy for using AI in our application process.