AI Safety and Alignment Jobs (2026)

AI safety is a small hub, and it pays to read the categories before the count. Of the 77 live listings as of September 2026, 35 are classified as AI Annotation / Data Labeling: contract work rating model outputs and flagging harmful content, which is safety work in the broad sense but not alignment research. Eight are Research Scientist roles, three are ML Engineer and two are NLP / LLM Engineer. Four arrived in the past seven days.

The research and engineering roles are where the titles get specific: staff research scientist for AI safety, senior engineering manager for an AI safety platform, senior software engineer for safety detection precision, and the Anthropic Fellows Program for AI safety and security. Handshake is the largest identifiable poster with 8 listings, then Anthropic with 4, Pinterest with 3, and Samsara, SpaceX and Biohub with 2 each. The US has 23 located roles, the UK 6 and Singapore 3.

Seventy percent of the hub is flagged remote, well above the 55 to 60 percent typical of the engineering hubs, largely because the annotation contracts are distributed by design. Levels skew junior: 58 mid level, 10 staff or principal, and 4 each at senior and lead. A US dollar figure appears on 12 percent of listings; their median midpoint is about $239,500, driven by the research roles rather than the annotation ones.

To find alignment and interpretability work, set Category to Research Scientist or ML Engineer, or search for interpretability, red teaming, evaluations or alignment. The Anthropic jobs company page lists every role from the lab, and the AI governance jobs hub covers the policy side. Browsing is free. Membership is $9 a year or $4.90 every three months and unlocks the apply link and the full description on every job.