OpenAI Pays $445K to Study Self-Improving AI Risks

OpenAI is offering a Senior Machine Learning Engineer role with up to $445,000 in total compensation. This position focuses on the risks of advanced AI systems, particularly self-improving behavior and automated learning loops. Here's the thing: this isn't a typical machine learning job. Instead of building more powerful models, the engineer will identify risks in frontier AI systems and run safety evaluations. They will study autonomous tool use and how systems behave during self-improvement or retraining loops. The role also involves building detection and containment systems for risky behavior. What's interesting is that the compensation reflects how specialized and high-stakes this area has become. The core idea is to understand recursive self-improvement, where AI systems improve themselves through repeated automated cycles. The job aims to detect early signs of risky automation and prepare controls. The bottom line: this role highlights a growing concern about keeping future AI systems safe, predictable, and under control as they become more autonomous.