10a Labs

10a Labs is a safety and threat-intelligence company that describes itself as "Intelligence for Artificial Intelligence." It serves as the trust, safety, and security layer for organizations deploying AI at scale: frontier AI labs, AI unicorns, Fortune 10 companies, and leading global technology platforms. Its core services are adversarial red teaming (stress-testing AI systems to find how they can be misused or made to fail), model evaluations, and multilingual threat-intelligence collection, combining proprietary technology with deep human expertise to detect abuse at scale and help engineering, safety, and security teams stay ahead of evolving threats. Because the work is about finding real-world abuse, harmful content, and adversarial behavior across languages and cultures, its multilingual and regional expertise is central rather than incidental. 10a Labs was founded by Robert L. McKenzie and is based in Washington, DC, with a small team in the 11 to 50 range and hiring managed through Greenhouse. The company does not publicly disclose funding details or a founding year, so treat those as unknown. For workers, 10a Labs is relevant on two fronts: full-time roles in AI safety, security, and threat intelligence, and specialist multilingual red-team and analyst work that can suit people with regional and language expertise in online harms.

10a Labs hires in two overlapping areas. (1) Full-time roles: AI safety and security researchers, red-team engineers, threat-intelligence analysts, machine-learning and evaluation engineers, trust-and-safety specialists, product, and operations. These are recruited through the company's Greenhouse board and are frequently remote-friendly, with a Washington, DC base. (2) Specialist and multilingual red-teamers and analysts: because the work spans online harms, abuse, and adversarial behavior across many languages and regions, 10a Labs values native or near-native fluency in specific languages plus domain knowledge of trust-and-safety issues (for example CSAM/harm taxonomies, extremism, fraud, or influence operations), which can suit experts who bring regional and linguistic depth. Across both, the strongest candidates combine genuine trust-and-safety or security expertise with comfort doing adversarial, detail-heavy work on frontier AI systems. Note that some of this content is sensitive and can be distressing, similar to content-moderation work, so candidates should be aware of the nature of the material.