Weekly AI Job Risk Summary
Week: May 6, 2026
This week’s AI job risk update is mostly stable, with only small relative changes across occupations. The biggest signals came from stronger enterprise AI adoption, faster hardware demand, new model-debugging tools, and several security and autonomy stories. Apple’s comment that AI adoption is happening faster than expected, plus discussion of operationalizing AI for scale and sovereignty, supports slightly higher risk for digital knowledge-work roles tied to content, analysis, support, and software production. Goodfire’s new mechanistic interpretability tool also modestly strengthens the outlook for more controllable and deployable models in coding and information-heavy workflows. At the same time, cybersecurity news—including NSA testing Anthropic tools, OpenAI’s advanced security mode, and broader warnings about cyber-insecurity in the AI era—reinforces demand for human security oversight, tempering “jobs AI will replace” narratives for some defense-oriented roles. Waymo complaints and ongoing limits in robotics and real-world autonomy continue to support lower near-term AI replacement odds for many physical, safety-critical, and hands-on jobs often seen as more AI-proof jobs.
This Week in Context
These paragraphs turn the weekly table into a readable explanation of where automation pressure broadened, narrowed, or stayed steady.
The weekly average risk moved upward, which suggests pressure broadened rather than staying isolated to only a few roles.
The clearest upward pressure appeared in Customer Support Representative, Customer Support, Software Tester. Moves like these often show where AI is taking on more repeatable drafting, comparison, coordination, or first-pass analytical work.
Relative pressure eased most in System Administrator, Robotics Engineer, Cybersecurity Analyst. That does not make these roles permanently safe, but it does suggest this week's signals were less aggressive than in the roles moving upward.
Urban Planner, Urban Farmer, Investment Banker changed little and help anchor the baseline for this week. Stable roles matter because they show where the ranking is holding its shape even while other parts of the market move.
Read these paragraphs together with the linked news and the full ranking. The point is not a one-week prediction of replacement, but a clearer view of where automation pressure is concentrating first.
How to Read This Week
This report works best when you read the summary, score movement, and linked news together. Treat it as a weekly reading of changing automation pressure, not as a one-week prediction that a profession will immediately disappear.
Weekly Average Risk
44.65
Week-over-week change
+0.10
Jobs moving up
22
Jobs moving down
3
Jobs unchanged
179
Trend
The chart shows how the overall weekly average has moved. It helps separate a broad market shift from changes limited to a smaller set of jobs.
Where Pressure Rose First
These jobs posted the strongest upward moves this week. Read them as signs of where automation pressure is tightening fastest right now.
Where Pressure Eased
These jobs moved downward this week. A lower score does not mean the role is safe forever, but it does suggest less immediate pressure relative to the prior week.
Roles That Stayed Relatively Steady
These jobs changed little this week and help anchor the broader picture. Stability often matters as much as movement when judging whether a shift is broad or narrow.
This Week Ranking
Use the full ranking as a current snapshot of relative pressure across jobs. The score alone matters less than the combination of score, week-over-week change, and the task mix behind the role.
AI News Used In This Weekly Evaluation
The articles below are the main signals used in this week's evaluation. Read them as context for why pressure rose, fell, or stayed stable.
MIT Technology Review / 2026-05-04
Wired / 2026-05-02
Wired / 2026-05-01
MIT Technology Review / 2026-05-01
MIT Technology Review / 2026-05-01
MIT Technology Review / 2026-05-01
Wired / 2026-05-01
MIT Technology Review / 2026-05-01
Wired / 2026-04-30
Wired / 2026-04-30
MIT Technology Review / 2026-04-30
Wired / 2026-04-30
Wired / 2026-04-30
Wired / 2026-04-30
Wired / 2026-04-30
Wired / 2026-04-29
Wired / 2026-04-29
Wired / 2026-04-29
Wired / 2026-04-29