Security: AI training data and RL environment vendors
Vendors with public evidence of security work. This is a shortlist by domain, not a quality ranking; multi-domain vendors appear here only when this specific domain is supported.
Security: AI training data and RL environment vendors
21 vendors. 6 state they have a SOC 2 Type I or Type II report and 4 state they can run in a customer's cloud or on-prem. These are sourced statements, not independent verification. Pending and unspecified SOC 2 claims are excluded.
— means not publicly listed or not confirmed; it never means zero or “No”. Checked is the date a company’s facts were last checked, and each fact on its profile links to its source.
| Compare | Vendor | Offerings | Domains | Data sources | Funding | Checked |
|---|---|---|---|---|---|---|
| AbundantAbundant builds reinforcement-learning environments, evaluation benchmarks, and training datasets for AI labs, AI startups, and large enterprises developing and improving AI agents. | RL envs, Training data, Evals | Coding, Computer use, Enterprise workflows, Security | Open data | $500K | 2026-09-29 | |
| AppenAppen supplies AI labs and product teams with human-produced training data, annotation, expert feedback, RL environments, and model evaluations for machine-learning and agentic AI systems. | RL envs, Human data, Evals, Training data, Annotation | Multi-domain, Medical, Legal, Science, Finance, Coding, Browser, Security, Enterprise workflows | Expert-made, Licensed data | — | 2026-10-08 | |
| ARIMLABSBuilds production-style simulated workspaces and verifiable benchmarks for AI agents, helping AI developers train and test long-horizon tasks in cybersecurity, incident response, software engineering, and tool use. | RL envs, Evals, Simulation | Multi-domain, Security, Coding | Expert-made | — | 2026-10-13 | |
| BugcrowdBugcrowd provides cybersecurity training environments and an exploitation benchmark to AI developers, alongside crowdsourced vulnerability testing services for organizations. | RL envs, Evals | Security | — | $230M | 2026-09-29 | |
| ClaruClaru provides custom and licensed video, robotics demonstrations, and human annotations to AI labs building robotics, embodied AI, and world models. | Training data, Human data, Annotation, Evals | Robotics and physical, Games, Creative, Multi-domain, Coding, Legal, Medical, Math, Science, Security, Other | Licensed data, Expert-made, Synthetic, Open data | $170M | 2026-10-13 | |
| CollinearCollinear sells stateful simulated environments, agent evaluations, and synthetic training data to frontier AI labs and enterprise AI teams building and improving AI agents. | RL envs, Training data, Evals | Multi-domain, Security, Finance, Enterprise workflows | Synthetic | $10M | 2026-09-29 | |
| DatacurveDatacurve sells expert-built coding data, reinforcement-learning environments, benchmarks, and evaluations to foundation-model labs and AI developer-tool companies. | RL envs, Training data, Evals, Human data | Coding, Science, Security | Expert-made | $17.7M | 2026-09-29 | |
| exponentialExponential sells AI developers multi-agent security environments for RL post-training, security-focused training data, and adaptive red-team evaluations that test AI models and agent systems against realistic attacks. | RL envs, Training data, Evals | Security | — | — | 2026-10-13 | |
| Gray Swan AIGray Swan sells AI-model and agent security to frontier labs and enterprises: adversarial testing before release, plus runtime filtering and monitoring to block attacks, unsafe outputs, and policy violations. | Evals | Security | Expert-made | $50M | 2026-09-29 | |
| IdlerIdler sells reinforcement-learning environments, benchmarks and evaluation datasets based on production work to AI labs and other teams training and evaluating models. | RL envs, Evals | Multi-domain, Enterprise workflows, Legal, Coding, Security | — | $9.5M | 2026-09-29 | |
| IncalmoIncalmo builds reinforcement-learning cyber environments and security evaluations for AI labs and security teams to train and assess cybersecurity-capable models. | RL envs, Evals | Security | Synthetic | $8.7M | 2026-09-29 | |
| InnodataInnodata provides AI developers and enterprises with training data, human feedback, annotation, reinforcement-learning environments, model evaluations, and red-teaming services for building and improving AI models and agents. | RL envs, Human data, Training data, Evals, Annotation | Coding, Computer use, Browser, Enterprise workflows, Finance, Legal, Medical, Science, Math, Security, Robotics and physical, Multi-domain | Customer data, Expert-made, Synthetic | — | 2026-10-13 | |
| LabelboxLabelbox sells simulated reinforcement-learning environments, expert-generated preference and training data, and evaluation services to frontier AI labs; it also offers enterprise teams software for running agents on real company workflows. | RL envs, Human data, Evals, Training data, Annotation, RL infra | Multi-domain, Coding, Computer use, Security, Enterprise workflows, Science, Robotics and physical, Finance, Legal | Licensed data | $189M | 2026-09-29 | |
| LatchLatchBio sells cloud data and compute software, bioinformatics workflows, and AI-based analysis to biotech R&D teams, laboratories, and scientific kit or instrument providers. It also publishes evaluations and offers assessment services for biology-focused AI agents. | Evals | Science, Security, Medical | Expert-made | $163M | 2026-09-29 | |
| NoumadNoumad builds formally verified reinforcement-learning environments for post-training AI agents on unsolved problems in long-horizon machine learning, computational science, robotics and cybersecurity. | RL envs | Science, Robotics and physical, Security, Other, Multi-domain | Synthetic | — | 2026-10-13 | |
| Olam LabsOlam Labs builds simulated multi-agent environments, behavioral evaluations and game-derived datasets for AI labs and researchers training or evaluating models’ social, agentic and safety capabilities. | RL envs, Training data, Evals, Simulation | Coding, Computer use, Enterprise workflows, Finance, Games, Security, Multi-domain | — | — | 2026-10-13 | |
| QuesmaQuesma publishes coding-agent benchmarks and builds RL training environments for frontier labs; it also markets tools that help development teams understand agent costs, workflows and outcomes. | RL envs, Evals | Coding, Security, Games | — | $2.3M | 2026-10-06 | |
| RL LabsRL Labs builds long-horizon software-engineering environments where frontier AI labs can train models on complex, multi-step coding tasks. | RL envs | Coding, Computer use, Security, Enterprise workflows | Synthetic | — | 2026-09-29 | |
| TELUS DigitalAcquired by TELUSTELUS Digital supplies AI labs and developers with expert-built training data, annotations, agentic RL environments, red-team testing and evaluations, plus custom multimodal and physical-AI datasets and managed data-collection services. | RL envs, Human data, Training data, Evals, Annotation | Coding, Enterprise workflows, Finance, Legal, Medical, Math, Robotics and physical, Science, Security, Other, Multi-domain | Expert-made, Customer data | — | 2026-10-13 | |
| Trajectory LabsTrajectory Labs runs safety evaluations, expert red-teaming and RL environments that help AI labs find model vulnerabilities and measure safer behavior. | RL envs, Evals, Human data | Multi-domain, Security | Expert-made | — | 2026-09-29 | |
| TuringTuring supplies AI labs with training data, evaluations, and RL environments for coding, enterprise knowledge work, and STEM. It also helps large enterprises build and deploy AI agents in important workflows. | RL envs, Training data, Evals, Human data | Multi-domain, Coding, Enterprise workflows, Science, Math, Finance, Legal, Medical, Computer use, Security | Customer data | $225M | 2026-10-13 |
No vendor matches every filter. Filters in different groups must all match. Remove one to widen the list:
— not publicly listed or not confirmed. Open a profile for each fact’s source. The register is unranked; sorting only reorders.
How to read domain and SOC 2 labels
Multi-domain means work across several task areas, not every area. Vendors appear in a specific domain filter only when that area has supporting evidence. SOC 2 entries are vendor statements, not independently verified audits: Unspecified means claimed without a stated type; Pending means in progress; Not publicly listed means we found no public statement. None of those means a completed Type I/II report. SOC 2 and deployment details remain on profiles and in comparison.