Astral Codex Ten

Apollo Research

3 min read

Apollo Research

Article

Apollo Research is a recurring organization in the Astral Codex Ten archive, appearing 2 times across 2 issues between November 26, 2025 and February 12, 2026. The archive places it in contexts such as “Currently, two nonprofits - METR and Apollo Research - do similar tests”; “Marius Hobbhahn of Apollo Research”. It most often appears alongside METR, 2010 kink, 2024 kink.

Metadata

  • Category: Organizations
  • Mention count: 2
  • Issue count: 2
  • First seen: November 26, 2025
  • Last seen: February 12, 2026

Appears In

Source Context

Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.

Why AI Safety Won't Make America Lose The Race With China
November 26, 2025 · Original source
These are relatively cheap asks. For example, the evaluation to see whether AIs can hack infrastructure will require hiring people who can conduct the evaluation, allocating compute to the evaluation, etc. But on the scale of an AI training run, the sums involved are tiny. Currently, two nonprofits - METR and Apollo Research - do similar tests on publicly-available models. I estimate their respective budgets at $5 million and $15 million per year. Nonprofits can always pay lower salaries than big companies, so it may cost more for OpenAI to replicate their work - for the sake of argument, $25 million. Meanwhile, the likely cost to train GPT-6 will probably be about $25 - $75 billion, with a b. So the safety testing might increase the total cost by 1/1000th. I asked some people who work in AI labs whether this seemed right; they said that most of the cost would be in complexity, personnel, and delay, and suggested an all-things-considered number ten times higher - 1% of training costs.
Inline links: $15 million, $25, $75
What Happened With Bio Anchors?
February 12, 2026 · Original source
This is why you do a sensitivity analysis, and Cotra did this at least in spirit (talked about which parameters were most important; gave people widgets they could use to play around with). But it didn’t work as well as she might have hoped, giving a <10% chance of timelines as short as the current median. Several later commenters and analysts had good takes here, especially Marius Hobbhahn of Apollo Research. Along with correctly guessing that algorithmic progress would go faster than Bio Anchors predicted (albeit with the benefit of two more years of data), he wrote that:
Inline links: Marius Hobbhahn