Which Jobs Are Most at Risk From AI? New Anthropic Data Offers Clues.

Amid uncertainty about whether artificial intelligence is disrupting the job market, Anthropic has set out to provide some much-needed clarity. The company recently released a new study aimed at developing effective methods for measuring AI’s impact on jobs, pooling data from its own economic index, the U.S. Department of Labor’s O*NET database and a 2024 study that gauges which tasks are most likely to be automated by large language models (LLMs). 

By gaining a better sense of how AI has affected existing roles and what tasks it’s likely to master, Anthropic hopes to predict how AI could reshape the job landscape moving forward. The study presents mixed findings, however, likely doing little to assuage tech workers’ anxieties as the federal government signals it’s unlikely to impose strong regulations on a technology that promises to upend everyday life. 

AI in the WorkplaceAI Is Creating a Generational Divide at Work — and It’s Growing.

What Does Anthropic’s Labor Market Study Say? 

According to Anthropic’s study, AI isn’t yet the job-wrecker some have feared, but there are signs this could change in the near future.  

Observed Exposure Could Precede Automation 

Anthropic explored a new variable it dubs “observed exposure,” which measures tasks that LLMs should be able to perform in theory and that are actually being automated in workplaces. Researchers reviewed usage data collected from the company’s Claude chatbot, considering whether Claude supported full automation versus augmentation efforts when gauging a task’s AI exposure, as well as the time spent on each task. Based on this approach, Anthropic determined the following roles to be the most exposed to AI-driven automation: 

• Computer programmers 
• Customer service representatives 
• Data entry workers  
• Medical record specialists 
• Market research analysts 
• Sales representatives 
• Financial analysts 
• Software QA analysts 
• Information security analysts 
• Computer user support specialists 

The results shouldn’t come as a surprise, given that these roles involve work that is data-heavy, repetitive and rule-intensive — all ideal conditions for AI models. In addition, upgrades to Anthropic’s Claude Code have fueled concerns that software jobs could soon become obsolete, so it’s easy to see why computer programming and data-related roles dominate this list. 

These findings confirm that computer science is under severe threat from AI, while sectors like sales and finance are likely to undergo further automation as LLMs begin to address more complex problems. Jobs that emphasize physical labor, including motorcycle mechanics, cooks and dishwashers, remain the least likely to be automated. 

AI Is Still Far From Reaching Its Full Potential 

At the same time, just because AI has a high probability of automating certain tasks doesn’t mean that it’s done so. In mathematics, finance, engineering and other areas most exposed to AI, Anthropic found that few tasks are actually being automated. In other words, workers aren’t widely applying AI tools to complete tasks that they’re well-suited for. 

This limited usage comes even as companies increasingly deploy AI agents — advanced systems that can execute complex, multi-step tasks on their own. A benchmark developed by Mercor researchers revealed that these agents still lack the reasoning, problem-solving and planning skills for real-world settings, however. Their performance can also be influenced by hallucinations and data-based errors, resulting in larger issues when teams of agents operate based on faulty data and produce a series of mistakes that must be corrected. 

Given the risks and potential headaches AI poses, workers may have adopted a more conservative approach. That said, the technology is slowly spreading throughout the workforce, and certain groups of workers are particularly vulnerable to automation.  

Tech’s Gender Divide Could Get Worse 

According to Anthropic’s findings, workers most at risk from AI technologies are more likely to be female and have at least some college education, specifically an associate’s or bachelor’s degree. This statistic reflects the troubling trend of women frequently being casualties in tech layoffs, while they already represent a smaller portion of the tech workforce. 

Part of the problem could be that women often come up against the “glass ceiling,” a metaphorical barrier preventing women and workers in other marginalized groups from advancing up the career ladder. As late as 2024, the U.S. Equal Employment Opportunity Commission found that “discrimination” is the main reason women and other marginalized workers struggle to gain a foothold in the tech sector. 

As a result, women tend to get stuck in the lower-level jobs that AI can more easily automate, rather than becoming managers or senior leaders who will likely be the ones orchestrating agents and other AI tools. In an industry plagued by weak DEI efforts, automation will cut workers who have been locked out of higher positions. And gender bias isn’t the only issue exacerbated by AI. 

Younger Workers Are Already Vulnerable 

Concerns for younger workers entering the labor market continue to rise amid shaky career prospects. According to the Anthropic study, hiring rates for occupations less exposed to AI have remained steady, while hiring rates for more exposed occupations have decreased by half a percentage point. This difference isn’t statistically significant, but it does suggest a stagnant job market for aspiring professionals seeking to break into tech. 

In response to an iffy tech market, the skilled trades are experiencing a resurgence. According to a BlackRock investor letter, the trades are expected to see a job growth rate of 5 percent between 2024 and 2034, outpacing the national average of 3 percent. Younger workers are spearheading this surge, with more students questioning the value of a college degree — especially a computer science degree — in the age of AI. 

Of course, it’s still too early for current and aspiring tech workers to panic. But rumblings among younger workers could be the first sign that AI is becoming a force to be reckoned with, highlighting the need for further research to follow up on Anthropic’s study and create a clearer picture of how exactly AI is affecting the job market, if at all. 

The Silver Lining to the AI-Driven EconomyWho Wants to Be an AI Millionaire?

Why Did Anthropic Conduct This Study Now? 

As more companies embrace agentic AI, the technology has reached a critical inflection point. Anthropic’s own Claude Code has contributed to fears of an impending “SaaSpocalypse” that could see the collapse of the traditional software-as-a-service business model, and it doesn’t help that organizations have begun to attribute layoffs to AI. That said, other research has found that AI may not be the main factor behind America’s current economic woes. 

These conflicting signals have only sown confusion around what could be driving a white-collar recession, encouraging tech workers to push back against AI in the workplace as part of a broader anti-AI movement. With AI companies boosting spending to accelerate AI adoption, Anthropic’s findings make it more urgent than ever to consider what an increasingly automated labor market means for workers. 

Why U.S. Policy Could Continue to Favor AI Over Tech WorkersThese Super PACs Are Plotting an AI Regulation Showdown in 2026

Should Tech Workers Be Worried About Their Jobs?  

While thoughtful regulation could ensure AI benefits the whole of society, the federal government doesn’t seem to be stepping in anytime soon. President Trump has reaffirmed the deregulatory stance of his AI Action Plan, signing an executive order to override state AI laws. And while he has introduced a national AI framework, it appears to encourage a relaxed approach to regulating the technology. Meanwhile, members of Congress are in the early stages of searching for answers about how AI is affecting the workforce. 

With the federal government providing almost no guidance, workers may need to take it upon themselves to prepare for a future where the labor market revolves around AI. They can stand out among their peers and job applicants by learning how to manage AI agents and collaborate with AI tools to complete various projects. Soft skills like communication and leadership will also become invaluable for overseeing agents as simple, repetitive tasks are automated. 

A doomsday scenario in which AI wipes out all jobs seems unlikely based on Anthropic’s research, but nothing can be taken off the table just yet. Given the AI industry’s unpredictable behavior amid a volatile political environment, tech workers may need to hope for the best while preparing for the worst. 

Frequently Asked Questions