Up to 666.5 million won in annualized salary… Anthropic seeks AI safety researchers

Anthropic has opened applications for a paid AI safety research fellowship targeting experienced technical professionals, offering full-time stints of six weeks to six months on research team projects. The program pays up to 6.665 billion won annualized (roughly $4.8 million at recent rates — actually about $4,800–$5,000 per week, or approximately $250,000 per year), with the company citing quantitative-field professors, technical founders, and senior engineers as ideal candidates. Notably, machine learning experience is not required, nor are PhDs or publication records, reflecting Anthropic's view that researchers from physics, mathematics, neuroscience, and software engineering can contribute to AI safety research. The move signals intensifying competition for AI safety talent as labs scale alignment work alongside frontier model development. Watch for whether rival labs launch similar cross-disciplinary recruiting programs and how Anthropic's safety research headcount grows ahead of its next model releases.