Top 100 AI-Safe Jobs
These are the 100 roles in the AI Career Index dataset with the lowest exposure scores: the careers best protected from being replaced by AI. The list uses the same 0 to 100 score used across the index, worked out the same way every time, flipped to show the lowest-exposure roles. Same role, same inputs, same ranking every time.
Why these jobs are safe from AI
The roles here last for one of four reasons. Some require a legal licence (medical, legal, engineering, accounting partner-track). Some carry personal responsibility that no software can take on (judges, executive officers, quality leaders, chartered professionals). Some need someone on site (proceduralists, surgeons, skilled trades, frontline operators). Some involve conflict only humans can handle (security investigators, regulators, complex deal-making). What they share is that the part a machine can do is small next to the human judgment that cannot be replaced.
AI-safe does not mean AI-free. Roles on this list still use AI to help with their tools: clinicians use ambient scribes, lawyers use Harvey, surgeons use AI-augmented imaging, executives use AI-drafted memos. The difference is that AI sits under the responsible human rather than replacing them. As AI takes over the work around these roles, the people who hold them tend to gain value, not lose it.
Use this list as a guide for school-leaver decisions, mid-career moves, parental-advice conversations, or planning work where safety from AI is part of the requirement. The Top 100 Best-Paid Durable Roles ranking takes this same low-risk group and re-sorts it by US median wage, which is the practical answer when both safety and pay matter.
Skip the leaderboards. Jump straight to your role.
Type your job title to open the full read on your role: risk band, how much AI can automate, durable skills, the six-month plan, and where the role sits in the wider ranking.
Top 100 Roles with the Lowest Risk of Automation
Reading the columns: Median Wage from BLS OEWS, May 2025. AI Adoption is the share of work in the role already showing real-world AI usage today, from the Anthropic Economic Index; a 0% reading means no measurable usage in the AEI sample, not that AI cannot apply. Econ. Impact is the role's total US economic value (median wage × total US employment). Score is the combined AI exposure score, from 0 to 100.
How these rankings are computed
Two things feed every row: a score we work out ourselves the same way every time, and three outside public datasets that sit alongside it without changing the score.
Exposure scores come from the mix of work in each role. Same role, same score, every time. No machine learning, no survey data.
Each role is split into routine, strategic, and relational work, using a category baseline plus role-specific tweaks. The routine share drives the exposure score.
US median annual wages from BLS OEWS. Wage compression multiplies the share of routine work by median wage to show where the dollar value of compression is largest.
Total US economic value of the role, computed as median wage multiplied by total US employment from BLS OEWS. Puts each role's ranking against the bigger economic picture.
Adoption is the share of work in a role that already shows real-world AI usage. It separates roles where AI could apply from roles where it already does. A 0.0% reading means the AEI sample observed no measurable usage for that occupation, common for specialised medical, legal, and trade roles whose practitioners use their own specialist software rather than general-purpose chat assistants.
Rankings update when roles are added or inputs change. See the full Methodology for the scoring inputs, the weighting, and the full set of rules.
Download the complete ranking
All 1592 roles with their exposure scores, work mix, median wage, and wage compression in CSV or JSON format. Available to Premium members and workforce-licensed teams.
One-time $49 unlock
Common questions about this ranking
Other rankings in this set
Three other views of the same data, each answering a different career-planning question.
What's holding through the AI economy
Articles on the traits that protect a role from being replaced by AI, plus the careers and industries gaining value in real terms.
- What Makes a Job Safe from AI Automation?AI-safe does not mean AI-free. Four structural moats protect careers from AI: licensure, named accountability, physical presence, and adversarial dynamics.
- What AI Cannot Replace: A Structural InventoryFive structural categories of work that AI cannot replace in 2026. The list is shorter than most articles claim and more durable than most careers assume.
- The 25 Most AI-Insulated Jobs in 2026The roles AI cannot easily touch: protected by licensure, physical presence, judgment, and accountability. Ranked and explained.
- Which Industries Are Safest from AI?Some industries carry structural insulation that others do not. This is the 2026 reading across healthcare, legal, construction, security, and the categories where human accountability is the moat.
- Jobs AI Cannot Replace: The Accountability MoatThe roles that stay durable against AI share one structural trait: accountability that must be borne by a human.
- How to Future-Proof Your Career Against AIA practical framework for identifying your structural exposure and moving toward durable work over the next 12 months.
- 10 AI-Proof Careers You Should ConsiderTen career paths that are structurally insulated against AI compression, ranked by durability and demand.
- Which Jobs Are Safe from AI in 2026?A structural look at which jobs remain durable against AI compression and the traits they share.
- The 5 Roles Gaining Authority in the AI EconomyWhile compression hollows out the middle, five categories of role are accumulating structural authority. They share traits worth understanding.
Measure Your Position with AI Career Index
Take the AI Career Snapshot to see your structural leverage, automation exposure, and growth trajectory with AI Career Index.