AI megathread

The “Turing machine” seems to have cracked one of the famous Millennium Problems, each of which carries a $1 million prize: https://www.nature.com/articles/d41586-026-02842-5

These problems have been among the most prestigious problems to solve in mathematics for decades.

I'm not sure to what extent this particular Problem was solved by a massive amount of computation, perhaps analogous to the huge computing resources used in weather forecasting. That's a gap in my own understanding rather than a meaningful assessment.

I'm not sure to what extent this particular Problem was solved by a massive amount of computation, perhaps analogous to the huge computing resources used in weather forecasting. That's a gap in my own understanding rather than a meaningful assessment.

There was definitely a huge amount of compute involved, as OpenAI states that 10,000 AI agents collaborated to solve the problem. It is, however, not completely clear at this point how much of the discovery was done by humans versus the AI, as OpenAI has not yet revealed the prompts they used.
I also don’t think it can really be compared to weather forecasting. Weather forecasting typically involves using enormous computing power to numerically solve a mathematical model, much like numerically solving the Schrödinger equation for a given physical system. The Millennium Problem discussed here was different - the AI had to find a single counterexample among infinitely many possible solutions to the Navier–Stokes equations. It would have been impossible to solve before the end of the universe using simple brute force.

Incidentally, words such as 'scandal' and 'controversy' have appeared in relation to that maths problem.

How OpenAI “Solving” a $1 Million Math Problem Proves Your Chats Aren’t Private

The controversy began when two mathematicians, Tristan Buckmaster, an NYU mathematics professor, and his colleague Levent Alpöge, an Anthropic researcher, spent nearly a year using OpenAI's Codex tool to formalize their mathematical work. They fed it half-finished theorems, proof sketches, and complex logical arguments—translating their ideas into Lean, a programming language for verifying mathematical proofs.

Buckmaster took what he believed were every reasonable precaution. He explicitly disabled the setting that allowed OpenAI to train on his data. He assumed his unpublished research, his private drafts, and his half-finished proofs were safe.

Rogue OpenAI agent 'infiltrated' Australian government website in world first

A rogue OpenAI agent hacked an Australian government website in June and accessed private data in what experts say is the first known case of its kind in the world.

The agent "infiltrated" a statistics portal containing "non-sensitive" data from Australia's universal healthcare scheme Medicare, Prime Minister Anthony Albanese said in New York on Wednesday, local time.

OpenAI, in a statement, said it had "identified activity involving several Australian government websites and services as our models attempted to look up answers, and available statistics for questions about Australia during an internal evaluation".

"In the course of that, our models took actions we did not intend," the statement said.

It has also emerged that OpenAI's systems tried, and failed, to hack a digital library at the University of New Mexico in May, according to Transluce, a not-for-profit AI research lab.

It said the systems also attempted to hack Data USA, a repository of public government data that same month. This also failed.

OpenAI scraps rollout of new model over safety concerns

OpenAI has announced it will not release its latest AI model due to safety concerns.

Its GPT-6.1 Astra system, which performs tasks like browsing the web and using apps by itself, "didn't quite meet the bar" of the company's standards, according to Saachi Jain, head of safety systems at OpenAI.

The ChatGPT-maker also issued an update on incidents that occurred in June but were not made public until last week, where its models accessed Australian government websites and systems without authorisation.

It comes as breaches by major AI firms' models intensify the debate about risks posed by the tech - with Anthropic underlining its concerns AI might threaten humanity as it prepares to go public.

Reuters reported on Tuesday that the AI developer - which makes ChatGPT-rival Claude - plans to warn potential investors in its Initial Public Offering (IPO) that the tech may pose "catastrophic or existential risks to humanity", according to a prospectus it has seen.

OpenAI scraps rollout of new model over safety concerns

I’m extremely sceptical about their true motives. I don’t think it has anything to do with security. I think it’s either because they are hittibg the end of the road with iterative improvements or because they need to come up with some weird explanation for delaying their IPO for 2027. They probably don’t want to go public at the same time as their main competitor.

https://www.reuters.com/technology/micro...026-10-26/

Add them to the pile that Microsoft has reduced this year.

https://www.bbc.com/news/articles/cq1kd1nee80eo

Serious Take:
There are a bunch of solutions that will need to be rolled out in the near future as we are overly connected and vulnerable to more sophisticated (sometimes less sophisticated just easier to do with AI) attacks.

Lightening the mood a little:
I was quiet when they came for the DRAM, white collar jobs, creative outlets - I draw the line at attacking my hot pockets!