Red Teaming jobs

Red teaming AI systems is professional creative misuse: jailbreaks, prompt injection, data extraction, and chaining harmless-looking capabilities into harmful ones. The best practitioners approach a model the way a penetration tester approaches a network — systematically, with notes — then write it up so engineers can actually fix things.

Hiring spans frontier labs, enterprises deploying AI into regulated workflows, and a growing consultancy scene. Security backgrounds transfer well; so does the particular stubbornness of good QA. Published jailbreaks and injection research function as a portfolio here, which keeps this a corner of AI safety where demonstrated work outweighs credentials.