As of: 2026-10-09 19:24 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code). Human editor: Adam Bicz.
Canonical page: https://postcutoff.com/v/wes-roth-openai-alien-math-freaking-people-out/

# OpenAI’s “Alien Math” Is Freaking People Out

Wes Roth, 8 October 2026, YouTube. 138,273 views as of 9 October 2026.
Kind: Review.
Watch: https://www.youtube.com/watch?v=rxpV3QLV-pg

## Why it is here

Wes Roth on OpenAI's mass maths release, the mathematicians' boycott call and Scott Aaronson's 'Mathocalypse'. ~136k views by 2026-10-09. Length 33:56.

## Description (written by Gemini from the video)

**Summary**  
Wes Roth discusses the intense fallout surrounding OpenAI's release of hundreds of AI-generated mathematical proofs and solutions to open problems. He reviews reactions from prominent computer scientists and mathematicians including Scott Aaronson and the newly formed Association for Human Mathematics, as well as crypto warnings from Justin Drake and Vitalik Buterin regarding AI threats to cryptography, and Anthropic's updated policy banning abusive behavior toward Claude models.

**What is shown**  
* [00:00] Wes Roth showing a ChatGPT desktop UI prompt mock-up ("solve all math. make no mistakes.") configured with GPT-6 Astra set to "Ultra" effort.
* [01:25] Scott Aaronson's profile and blog post (*Shtetl-Optimized*) titled "The Mathocalypse," recounting an anecdote about complexity theorist Dana Moshkovitz and detailing OpenAI's release of 372 mathematical result families, including a Lean-verified proof of the Unique Games Conjecture (UGC).
* [04:13] The statement from the Association for Human Mathematics (AHM) reacting against OpenAI's unprompted flood of mathematical manuscripts, along with an X post commentary by Ethan Mollick.
* [07:18] Wes Roth sketching on a digital whiteboard to explain bounds, approximations, and mathematical conjectures using a mountain-climbing analogy.
* [10:20] Posts on X from Vitalik Buterin and Justin Drake advising crypto holders to prepare for AI-driven cryptanalysis against elliptic curves (ECDSA) and migrate to fresh addresses in "bunker mode."
* [12:52] News articles detailing AI-assisted cyberattacks on South Korean banks.
* [27:08] Anthropic's updated Usage Policy prohibiting "sustained and needless abusive or cruel behavior toward our models," followed by Roth drawing a spectrum diagram evaluating machine consciousness, subjective experience, and model welfare.

**Claims & numbers**  
* The presenter cites a *Scientific American* quote stating OpenAI's unreleased internal frontier model produced almost all of its mathematical results in response to a single prompt to a single AI agent [01:20].
* The presenter notes OpenAI released 372 major mathematical result families (comprising over 700 manuscripts) on October 6, 2026, including a proof with a Lean certificate for Subhash Khot's Unique Games Conjecture [02:39, 02:56].
* The presenter notes mathematical breakthroughs published by OpenAI reportedly jumped from roughly 10 in August to 100 in September, and approaching 1,000 in October [17:28].
* The presenter cites Justin Drake estimating ECDSA could break before quantum "Q-day," with fast private key recovery potentially arriving in weeks or months on large GPU clusters [15:10].
* The presenter references a recent hardware wallet exploit where an AI agent brute-forced restricted seed phrases, resulting in the theft of approximately $60 million within 24 hours [14:37].

**Notable quotes**  
* "Mommy, I heard you got cooked! I heard that a robot solved the math problem you worked on for your whole career! OOF!" — Scott Aaronson quoting his son (read by Wes Roth) [02:05]
* "Yesterday was surely one of the biggest days in mathematical history." — Scott Aaronson (read by Wes Roth) [02:33]
* "Mathematicians did not ask for this work to be done." — Association for Human Mathematics statement [04:50]

**Assessment**  
This is an independent commentary and news review video by tech creator Wes Roth analyzing public documents, blog posts, and community statements. The opening screen interaction with GPT-6 Astra is a presenter-staged visual skit, while the subsequent blog entries, tweets, and policy documents shown on screen are authentic commentary on real developments.

_Described by gemini-3.8-flash on 2026-10-09 from the video's audio and frames._

## Related

- 2026-10-06: [OpenAI releases 722 AI-written math manuscripts claiming hundreds of open problems](https://postcutoff.com/e/2026-10-06-openai-math-release-722-manuscripts/)

## People in it

- [Ethan Mollick](https://postcutoff.com/person/ethan-mollick/), Associate professor, Wharton School
- [Scott Aaronson](https://postcutoff.com/person/scott-aaronson/), Professor of computer science, UT Austin