An Anthropic Co Founder Says the Singularity Has Begun

He said it to the people who run the global economy

Anthropic has spent 2026 warning that AI is moving faster than anyone is ready for. Now one of its co founders has given the moment a name, in front of the most establishment audience imaginable. The Rundown reports that Jack Clark told economists and policymakers at the IMF and World Bank annual meetings that the singularity has now begun.

His definition, shown on his own slide, is plain: the point when AI systems speed up the pace of overall scientific and technical progress in a way that is hard for humans to comprehend. This was not a podcast hot take. It was a co founder of a leading AI lab saying it on the record to the institutions that manage money and policy for the whole world.

The score jumps he used as evidence

Clark pointed to test results that are hard to wave away, even allowing for the fact that benchmarks are an imperfect measure of real ability:

• Scores on Humanity Last Exam, a set of expert level questions, rose from 8 percent to 67.7 percent since 2025
• FrontierMath performance improved from 2 percent to 94 percent over the same period
• One Anthropic model puts United States GDP at 44 trillion dollars by 2030 if AI spreads fast, a figure Clark admitted sounds outlandish
• He predicted very disruptive job impacts alongside that growth, and said the world 10 million scientists may soon direct a billion virtual scientists

He also talked about a ceiling on intelligence

The most striking part may be what Clark said about limits. He told the audience that chief executive Dario Amodei has floated limits on AI ceiling of intelligence, and that he believes there is only a small number of years left to act on questions like that. He also urged nations to build labs and robots before they run short.

Read that carefully. A leader inside one of the top labs is publicly discussing whether the technology should have a cap, while his own company races to build more of it. Whether you find that reassuring or alarming, it tells you the people closest to the work no longer treat runaway progress as a distant thought experiment.

Why this matters outside the IMF room

Singularity talk used to live in science fiction and in fringe corners of the internet. When it is delivered at the IMF, it becomes planning input for governments and big employers. Budgets, training programs, and hiring plans get shaped by what leaders in that room believe is coming, whether or not the boldest version arrives on schedule.

For small studios and creators, the practical translation is less dramatic and more useful. Tools will keep getting better at the draft stage of almost every job: research, first cuts, captions, layouts, and code. The scarce skills move toward judgment, taste, client trust, and finishing work to a standard a person will sign their name to.

Hold the claim and the caution at the same time

Clark has every incentive to make his industry sound world changing, and exam scores are not the same thing as scientific discovery in the wild. A healthy reader keeps both facts in view. The claim is promotional, and the trend underneath it is real enough that finance ministers are being briefed on it.

The useful response is not panic and not dismissal. It is to build your work so that each new wave of capability makes you faster instead of obsolete: own the client relationship, own the final call on quality, and let the machines carry the volume. If Clark is even half right about the pace, that positioning is the safest seat in the room.