AI Job Risk Index

Weekly Evaluation Summary

This week’s AI job risk update is mostly stable, with only small relative moves across occupations. The biggest signals came from continued progress in AI agents, including Microsoft’s new OpenClaw-like work on task-completing agents and strong enterprise momentum around Claude and AI-powered developer tools. That slightly raises exposure for jobs built around repeatable digital workflows, software support, and structured knowledge work. At the same time, cybersecurity and governance-heavy roles look a bit more resilient because Anthropic’s Mythos-related security concerns, facial-recognition backlash, and new lawsuits around harmful AI use all reinforce the need for human oversight, compliance, and incident response. Early humanoid robot commercialization from Unitree is notable, but it still looks too immature to materially change near-term replacement odds for most physical jobs. Overall, the latest pattern in jobs AI will replace and jobs at risk from AI still favors higher pressure on clerical, support, and routine content tasks, while many AI-proof jobs remain those requiring trust, dexterity, regulation, or high-stakes human judgment.