Is AI Quietly Closing the Door on Your Career? What Anthropic's New Data Actually Tells Us

*By the LinkedOut Editorial Team · May 25, 2026 · 8 min read* --- If you've been laid off recently — or quietly passed over for a role you were perfectly qualified for — there's a good chance you've wondered whether AI had anything to do with it. That question just got a more honest answer than most researchers have been willing to give. In March 2026, Anthropic — the AI safety company behind the Claude AI assistant — published a landmark research paper titled *"Labor Market Impacts of AI: A New Measure and Early Evidence."* Written by economists Maxim Massenkoff and Peter McCrory, it introduces a new way of measuring AI's real impact on jobs — not what AI *could theoretically* do, but what it is *actually doing right now* in professional settings. The findings are equal parts reassuring and deeply sobering. Here's what the research says, what it means for your job, and what LinkedOut thinks every worker needs to know. --- ## The Problem With Most "AI Will Take Your Job" Research Before getting into the findings, it's worth understanding why this paper matters methodologically. Most previous research on AI and jobs relied on *theoretical exposure* — essentially asking, "Could AI perform this task?" That approach led to headlines like "47% of US jobs are at risk of automation," published years ago by Oxford researchers, which turned out to be wildly overstated. Anthropic's team took a different approach. They built a measure called **Observed Exposure** — combining what AI *can theoretically do* with what people are *actually using it for in real professional workflows*, as measured through millions of real Claude conversations in the Anthropic Economic Index. The result: a much more grounded, conservative, and honest picture of where AI disruption is actually landing — right now, not in some hypothetical future. --- ## The Most Important Finding: AI Is Way Behind Its Own Potential Here's the headline number that should reframe how you think about this entire topic: **AI is currently covering only 33% of tasks in Computer and Math occupations** — even though theoretical models suggest it could handle up to 94% of those tasks. That is a massive gap. The blue area on Anthropic's Figure 2 (theoretical capability) dwarfs the red area (actual usage) across every job category. For Office and Administrative roles, theoretical exposure is 90%, but actual coverage remains far lower. For Healthcare, Legal, and Education, the gap is even wider. What does this mean for workers? It means that for the vast majority of professions, **AI is not yet at the door — it's still in the driveway.** The disruption is coming, but it hasn't arrived at scale. This is both a warning and a window of opportunity. --- ## Who Is Most Exposed Right Now? The paper identifies the **10 most AI-exposed occupations** using the Observed Exposure metric. These are the jobs where AI tools are already automating real tasks in real professional environments: 1. **Computer Programmers** — 75% task coverage. Coding is the single most automated profession in Claude's usage data, driven by code completion, debugging, and generation workflows. 2. **Customer Service Representatives** — High exposure from first-party API traffic, where businesses have deployed Claude to handle customer queries directly. 3. **Data Entry Keyers** — 67% coverage. Reading source documents and entering data is highly automatable and already heavily automated. 4. **Market Research Analysts** — Summarization, synthesis, and competitive analysis are heavily used AI tasks. 5. **Financial Analysts** — Report generation, data interpretation, and financial modeling are all seeing automation. 6. **Sales Representatives** — Outreach, personalization, and follow-up drafting are being automated. 7. **Quality Assurance Testers** — Code and content testing workflows have high AI penetration. 8. **Information Security Analysts** — Threat pattern analysis and reporting see heavy Claude usage. 9. **IT Support Specialists** — Troubleshooting scripts and technical documentation are prime AI territory. 10. **Medical Records Specialists** — Clinical documentation and record summarization are increasingly AI-assisted. Notice the pattern: **these are not low-wage, low-skill jobs.** These are educated, credentialed, often well-paid knowledge workers — exactly the demographic that was told for decades that higher education would protect them from automation. --- ## The Profile of an At-Risk Worker Will Surprise You One of the most striking findings in the research concerns *who* is most exposed. Workers in the top quartile of AI exposure — compared to those with zero exposure — are: - **16 percentage points more likely to be female** - **11 percentage points more likely to be white** - **Almost twice as likely to be Asian** - **47% higher earners on average** - **Far more educated** — graduate degree holders make up 17.4% of the most exposed group, versus just 4.5% of the unexposed group Let that sink …