Domain on the map

Cybersecurity benchmarks for LLMs and cyber RL environments

Offensive and defensive security work: finding and exploiting vulnerabilities, hardening systems and running penetration tests against sandboxed targets.

RL environments
4
AI benchmarks
2
Shares entries with
4 categories

Snapshot 2026-11-05. How categories are built · Where the data comes from

The galaxy could not load. See every vendor on the board.

Security, live from the universe. Hover a star for its name; click to open it. Open it in the universe

In short

What are RL environments for cybersecurity?
A simulated workplace where an AI agent practises a task (a code repository, a browser, business software) and an automatic grader scores each attempt. Labs train on thousands of these attempts. Bought to improve models. For cybersecurity, that means offensive and defensive security work: finding and exploiting vulnerabilities, hardening systems and running penetration tests against sandboxed targets.
Who builds RL environments for cybersecurity?
Collinear, Quesma, ARIMLABS and Incalmo.
What are the main cybersecurity benchmarks?
CWE-bench and BinaryAudit.

RL environments 4

  1. Collinear
  2. Quesma
  3. ARIMLABS
  4. Incalmo

AI benchmarks 2

  1. CWE-bench
  2. BinaryAudit

Data 6

  1. Gray Swan AIbased here
  2. Trajectory Labsbased here
  3. Collinearalso covers it
  4. Datacurvealso covers it
  5. ReasonCorealso covers it
  6. Reveloalso covers it

Domains that overlap with Security

Next: pick another domain. Tags come from the RL Research daily scrape, canonicalized nightly.