Alibaba's RSI announcement didn't get the attention it deserves. "Alibaba has revealed progress in RSI (Recursive Self-Improvement) driven by empirical feedback. Over a month of fully automated runs - spanning pipeline design, data validation, iterative experimentation, and error diagnosis - Qwen3.8-Max completed 33 iterative cycles. Through autonomous training optimisation and post-training techniques, the updated Qwen3.8-Max boosted its Artificial Analysis score from 40 to 45." From the press release: alibabacloud.com/en/press-ro…

To secure frontier AI access, Australia’s best bet is a coalition of the dependent | Hannah Merchant | bit.ly/4xJADW6

To all those saying duh just stop if you don’t think it’s safe … there are a few things riding on it … wsj.com/economy/the-ai-build…

“Feel,” even “want,” I can get behind as things for which evidence is thin. But AIs *clearly* think and understand. If your definition of “think” excludes entities that can autonomously resolve Millennium Problems, it is your definition that is flawed. This desire to avoid anthropomorphic language has become a pathology.

Ukraine-Australia Bilateral Security Agreement signed. I am deeply grateful to @AlboMP @RichardMarlesMP @SenatorWong for their support 🇺🇦🇦🇺💪

Today we have 3 new DeepMind Institute essays: How can we control misbehaviour in agent swarms? How should we orchestrate complex networks of AIs and people? The case for making AGI’s benefits equitably distributed. Get them here -> bit.ly/deepmind-institute

It was a pleasure meeting in Canberra—thanks for your leadership on this, Senator!

Always good to talk to LJ. It's a very interesting case that we need to learn from but it's the trend that we should really worry about more than this individual incident. Another alarm bell that should spur action.

I spoke with @SBSNews about the @OpenAI Medicare hack. We should be thinking of this as similar to an industrial safety issue. We need immediate incident reporting obligations, duties of care for company officers, and licensing for high-risk research. sbs.com.au/news/article/open…

Not in the least bit surprised by this.

I went back on The Diary of the CEO @StevenBartlett, this time for The Great AI Debate (@So8res, @edzitron, @amcafee). If you have not seen it yet, check it out!

CNN Solidarity. Reporting real news from the UN. @kylieatwood @kaitlancollins

I’m glad to see more people talking about the future of AI, both within the tech industry and outside of it. nytimes.com/2026/09/20/opini…

The Baltic Musketeers are together again. On a joint visit to Ottawa, we’re once again working to support Ukraine and secure victory together. 🇱🇹🇪🇪🇱🇻🇺🇦

The most powerful technology on earth should not be less regulated than a carton of milk. Great to chat with Microsoft President Brad Smith on why AI safety will ultimately require the right balance of innovation, industry and government regulation.

My fave moment from last week's Joint Select Committee hearing on AI. @DavidPocock with "And what is this leverage you speak of?" It's a fair question that many of us have been exploring. But just delightfully expressed here.

Great to see this cross-party call for Australia to act to address catastrophic AI risk. We recently briefed WA MPs on AI capabilities and risks, and policy options for Australia. Thanks to @liamstaltari and Tim Clifford for the invite. 1/4 theaustralian.com.au/nation/…

The most disingenuous takes going round are those saying, "The labs should just align AI." It's their problem and they should just get on and fix it. As if a frontier model that can solve Millennium Prize problems is a normal bit of code. Tendentious codswallop.