# Anthropic reports robots’ reach across work—and the barriers to adoption [Canonical reading](https://escapemeridian.com/readings/anthropic-reports-robots-reach-across-work-and-the-barriers-to-1aea85a7) Anthropic’s report, published September 30, 2026, argues that measuring what robots can do in jobs today may help identify where work could change first. Its robot exposure index rates tasks by whether robots can perform them and how structured the work setting must be. Anthropic estimates that robots can perform 74% of physical tasks in the US in some circumstances, but says they are cost-competitive with human labor for only 0.3% of job tasks. Most robots still need controlled environments; capability gaps, human preferences and regulations also limit adoption. The report says workers in more robot-exposed jobs are more likely to be male, less educated and lower paid, and that jobs more exposed to robots over the past 50 years saw greater declines in wages and employment than other jobs. These are exposure estimates and historical associations, not a forecast that particular workers will lose their jobs. Anthropic’s cost projections depend on robot prices following past trends, while its scenarios set aside factors including preferences and regulation. The report concludes that driving and warehouse jobs may change sooner than nursing and mechanics if the past is any guide, but says much remains uncertain about how robots and AI could reshape the economy. It leaves open how work itself may change as these technologies advance. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source Anthropic · Article Original source date: Sep 30, 2026 (2026-09-30T00:00:00.000Z). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-05T11:43:23.625Z. First observed: 2026-10-18T06:01:55.375Z. Retrieved: 2026-10-30T03:33:49.881Z. - [Can we predict the jobs robots will do?](https://www.anthropic.com/research/what-work-can-robots-do) ## Limits Text as retrieved on October 30, 2026; originally published September 30, 2026. The report describes its exposure measure as a way to assess current robot capabilities and relate them to possible future job disruption; it does not establish that exposure will result in job loss. The report says its robot cost estimates are approximate and that its scenarios apply the same cost decline or capability increase to every task. ## Open question How well will Anthropic’s current robot exposure index identify which jobs change first if robot capabilities or costs develop differently from the report’s assumptions? ## Source-stated Anthropic says its robot exposure index measures how well robots can perform job tasks today, including the kind of environment needed. > We present a robot exposure index based on how well robots can perform job tasks today. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic estimates that robots can perform 74% of physical tasks in the US in some circumstances, while cost-competitive robots can perform only 0.3% of job tasks. > Though robots can theoretically do tasks that add up to 34% of all work time, they are cost-competitive for just 0.3% today. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic reports that jobs more exposed to robots over the past 50 years experienced greater declines in wages and employment than other jobs. > Over the past 50 years, jobs more exposed to robots experienced greater declines in wages and employment than others. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic says that if the past is any guide, taxi drivers and warehouse packers will see changes sooner than nurses and mechanics. > If the past is any guide, taxi drivers and warehouse packers will see changes sooner than nurses and mechanics. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic says its adoption scenarios leave out forces such as preferences and regulation, making job impacts less certain. > Even so, job impacts are less certain with these cost projections, since they set aside forces like preferences and regulation as well as feedback effects like falling wages. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source and contributor context Anthropic · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Anthropic Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems. - [Explore this context](/entities/anthropic) - [Reference](https://www.anthropic.com/news) - [Reference](https://www.anthropic.com/research/team/economics) ## Editorial provenance Version 1. Publication ID: 58dbebef-ae63-0f45-5c31-cb1990620b23. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-05T11:38:49.277Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 3f011fb51577326bc2fdd99200eb23de6df455eded2c8f88db2c991b0d9eabf2. Source text hash: f6eca4d4505a4813917f314dce96e2d1441c6a2d66d8467a144f62faadd182ae. ## Revision and source-check history ### Version 1 — 2026-11-05T11:43:23.625Z Anthropic’s report, published September 30, 2026, argues that measuring what robots can do in jobs today may help identify where work could change first. Its robot exposure index rates tasks by whether robots can perform them and how structured the work setting must be. Anthropic estimates that robots can perform 74% of physical tasks in the US in some circumstances, but says they are cost-competitive with human labor for only 0.3% of job tasks. Most robots still need controlled environments; capability gaps, human preferences and regulations also limit adoption. The report says workers in more robot-exposed jobs are more likely to be male, less educated and lower paid, and that jobs more exposed to robots over the past 50 years saw greater declines in wages and employment than other jobs. These are exposure estimates and historical associations, not a forecast that particular workers will lose their jobs. Anthropic’s cost projections depend on robot prices following past trends, while its scenarios set aside factors including preferences and regulation. The report concludes that driving and warehouse jobs may change sooner than nursing and mechanics if the past is any guide, but says much remains uncertain about how robots and AI could reshape the economy. It leaves open how work itself may change as these technologies advance. ## Original source Anthropic · Article Original source date: Sep 30, 2026 (2026-09-30T00:00:00.000Z). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-05T11:43:23.625Z. First observed: 2026-10-18T06:01:55.375Z. Retrieved: 2026-10-30T03:33:49.881Z. - [Can we predict the jobs robots will do?](https://www.anthropic.com/research/what-work-can-robots-do) ## Limits Text as retrieved on October 30, 2026; originally published September 30, 2026. The report describes its exposure measure as a way to assess current robot capabilities and relate them to possible future job disruption; it does not establish that exposure will result in job loss. The report says its robot cost estimates are approximate and that its scenarios apply the same cost decline or capability increase to every task. ## Open question How well will Anthropic’s current robot exposure index identify which jobs change first if robot capabilities or costs develop differently from the report’s assumptions? ## Source-stated Anthropic says its robot exposure index measures how well robots can perform job tasks today, including the kind of environment needed. > We present a robot exposure index based on how well robots can perform job tasks today. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic estimates that robots can perform 74% of physical tasks in the US in some circumstances, while cost-competitive robots can perform only 0.3% of job tasks. > Though robots can theoretically do tasks that add up to 34% of all work time, they are cost-competitive for just 0.3% today. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic reports that jobs more exposed to robots over the past 50 years experienced greater declines in wages and employment than other jobs. > Over the past 50 years, jobs more exposed to robots experienced greater declines in wages and employment than others. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic says that if the past is any guide, taxi drivers and warehouse packers will see changes sooner than nurses and mechanics. > If the past is any guide, taxi drivers and warehouse packers will see changes sooner than nurses and mechanics. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source-stated Anthropic says its adoption scenarios leave out forces such as preferences and regulation, making job impacts less certain. > Even so, job impacts are less certain with these cost projections, since they set aside forces like preferences and regulation as well as feedback effects like falling wages. - [Supporting source](https://www.anthropic.com/research/what-work-can-robots-do) ## Source and contributor context Anthropic · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## Anthropic Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems. - [Explore this context](/entities/anthropic) - [Reference](https://www.anthropic.com/news) - [Reference](https://www.anthropic.com/research/team/economics) ## Editorial provenance Version 1. Publication ID: 58dbebef-ae63-0f45-5c31-cb1990620b23. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: passed by anthropic/claude-opus-5.5 on 2026-11-05T11:38:49.277Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model. Observation: 3f011fb51577326bc2fdd99200eb23de6df455eded2c8f88db2c991b0d9eabf2. Source text hash: f6eca4d4505a4813917f314dce96e2d1441c6a2d66d8467a144f62faadd182ae. ### passed — 2026-11-05T11:38:49.277Z — anthropic/claude-opus-5.5 Publication: 58dbebef-ae63-0f45-5c31-cb1990620b23 attribution — pass: The reading attributes the index, estimates, backtest and expectations to Anthropic's report, which matches the publisher page. The source is a single research report with its own argument, not a listing. No individual authors are named in the retrieved text, and the reading does not invent any. > We develop a measure of job exposure to robots, using Claude to assess how well present-day robots can perform work tasks. summary — pass: The summary leads with the actor and the argument that today's capabilities can show where work may change first. It explains the robot exposure index in plain terms and keeps the core tension between broad capability (74% of physical tasks) and narrow cost-competitiveness (0.3%). It also preserves the source's uncertainty and conditional framing. One small omission is that Claude produced the ratings, but this does not block understanding. The question follows from the source's own caveats. > Our core premise is that jobs are more likely to be impacted when robots can already do their work today. date — pass: The stated date, September 30, 2026, matches the originalPublished instant as a UTC calendar day. The page was modified on 2026-10-23. However, the cited passages are the report's core findings and are plainly integral to the original piece. Cited references date no later than September 2026. Forecasts such as the 40-year horizon and the taxi-versus-nurse ordering are kept as attributed, conditional expectations. There is no hindsight framing. The kind is publisher-page, so no archive disclosure is needed. > If robot price declines follow past trends, it will take 40 years for that share to reach 10%. support — pass: Each claim is source-stated: 74% of physical tasks in some circumstances, 0.3% cost-competitive, the 50-year association with wage and employment declines, the conditional ordering of affected jobs, demographics, and that scenarios set aside preferences and regulation. The reading's caveat that exposure is not a forecast of job loss is consistent with the source's own hedging and cost-parity discussion. > Counting any exposure level, 74% of physical work, or 34% of all work, can be done by robots in some circumstances. media — pass: No media was selected, and no verified candidate media were supplied, so there is no identity or relevance issue. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/1aea85a760e308a5503f483b44dd690abfbf4bda9ca9ca7f46d049a631dcd8d9/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/anthropic-reports-robots-reach-across-work-and-the-barriers-to-1aea85a7.json)