A safeguards researcher at Anthropic said nearly everyone will lose the ability to work. In a Palisade Research interview published November 15, 2026, Tao Lin said: “I think that it’s very likely that pretty much all humans on Earth will become permanently unemployable and will need to strictly rely on government welfare to sustain themselves, and human employment will go away.”
A clip of that answer, posted on X on November 17, had about 3.8 million views by 06:20 UTC on November 19. The full interview runs 1 hour 2 minutes and had 85,666 views on YouTube. Below is what he said around the clip, with times from the video.
On this page · 6 sectionsOpen
Tao Lin, who works on Safeguards at Anthropic, told Palisade Research it is very likely that pretty much all humans will become permanently unemployable and rely on government welfare.
He gave a timeline of about 1.3 years, or around the end of 2027, to broadly superhuman AI: AI as good as or better than the best expert in almost every field.
Asked about timing, he said he expects a large gap between AI being able to replace people and actually doing it, partly because there are not enough robots yet. The viral clip leaves that out.
He said AI companies should slow their research to something like a quarter of its current rate.
Asked whether his view was personal, he said it is quite common in the AI safety community and at Anthropic. Anthropic has not commented on the clip.
§ 01What he said
| Time | Topic | What he said |
|---|---|---|
| 00:42 | Who he is | Works at Anthropic about a year and a half; now on Safeguards, after automated alignment research and AI control |
| 03:58 | Timeline | About 1.3 years to broadly superhuman AI, around the end of 2027 |
| 11:07 | Whose view | Quite common in the AI safety community and at Anthropic |
| 15:33 | Path | People lose current jobs, move briefly to other work, then become unemployable |
| 15:58 | The clip | Pretty much all humans permanently unemployable, relying on government welfare |
| 16:21 | Timing caveat | A large gap between AI being able to replace people and actually doing it |
| 23:46 | Remedy | AI companies should slow research to about a quarter of today’s rate |
His timeline is for capability, not for job losses. He defined it as “Timelines to AI that is as good as or better than the best expert at almost every field.”
§ 02The part the clip leaves out
The clip ends on “human employment will go away.” Asked next about timing, Lin said he expects a large gap between when AI can replace people in all roles and when it actually does, partly because of how fast robots can be built: “There aren’t enough robots yet”.
Just before the clip, he described the path he expects: “everyone losing their current jobs, everyone needing to move to a different occupation briefly, and then sometime after that, being completely unemployable”.
Asked whether this was his personal view, he answered: “Yeah, I think that this view is quite common in like the AI safety community, also at Anthropic.” He did not say he speaks for the company, and the series opens with a card: “This collection of interviews is not representative of every perspective within the AI companies.”
His remedy is a slowdown: “I think that the AI companies should slow down.” He put a number on it, research at “something like 1/4 of its current rate”, and urged people to call their representatives.
§ 03How far it spread
| Date (UTC) | Where | What it was | Views |
|---|---|---|---|
| Nov 15, 23:25 | Palisade Research, YouTube | The full interview, 1:02:53 | 85,666 |
| Nov 17, 13:09 | @jessexai_ on X | A clip of the unemployability answer | 3,835,642 |
| Nov 18, 23:02 | @Polymarket on X | A one-line summary of the claim | 743,285 |
The clip outran the interview by about 45 to 1. Most people who heard the claim did not hear the caveat that followed it.
§ 04What Anthropic has said
We found no statement from Anthropic or from Lin about the clip as of 06:20 UTC on November 19. It landed the week Anthropic priced its IPO and a coalition published the MisAnthropic Statement, and it follows another Anthropic researcher’s resignation over self-improving AI.
§ 05Where CellCog sits
CellCog runs on Anthropic’s models and sells AI employees, so we have a stake on both sides of this argument. This page reports what Lin said, in his words and in order, and takes no side on his forecast.
§ 06Sources
- Palisade Research, interview with Tao Lin, YouTube, published 23:25 UTC, November 15, 2026
- @jessexai_ on X, the clip, 13:09 UTC, November 17, 2026
- Polymarket on X, summary post, 23:02 UTC, November 18, 2026
- frominside.ai, Palisade Research’s interview series, linked
Q1What exactly did Tao Lin say?
That it is very likely pretty much all humans on Earth will become permanently unemployable, will need to rely strictly on government welfare, and that human employment will go away.
Q2When does he think this will happen?
He gave about 1.3 years to broadly superhuman AI, around the end of 2027, but said he expects a large gap between AI being able to replace people and actually replacing them, partly because there are not enough robots yet.
Q3Is this Anthropic's official position?
No statement from Anthropic says so. Lin said the view is quite common in the AI safety community and at Anthropic; the interview series opens with a card saying it is not representative of every perspective inside AI companies.
Q4What does he think should be done?
He said AI companies should dramatically slow their AI research, to something like a quarter of the current rate, and that people should call their representatives and support candidates who want to slow AI down.
Q5Does CellCog use Anthropic's models?
Yes. CellCog runs Claude Opus 5.5 and Claude Fable 5.2 on its tiers, and it sells AI employees, so we have a stake on both sides of this debate. This page reports what Lin said and takes no side.
