Morning Briefing
An OpenAI safety leader has walked out the door, calling the company's culture "broken" — and it's not an isolated incident. As AI labs race to ship ever-more-powerful models, the exodus of safety-focused insiders is becoming a pattern that the industry can no longer dismiss as sour grapes.
What Matters Today
- OpenAI's safety brain drain continues: David Robinson has quit OpenAI, warning the culture is broken and that AI companies aren't being "nearly careful enough." This follows a string of high-profile departures. At the same time, Anthropic apparently tried to lobby the Pope on AI consciousness — make of that what you will. Guardian AU
- Flydubai co-pilot attacked captain with axe mid-flight: UAE authorities say the co-pilot used a crash axe to attempt a terrorist takeover of an Israel-bound jet. The suspect had reportedly been barred from flying by Oman over extremist views. A deeply alarming aviation security failure. BBC World
- Russia strikes second major Kyiv bridge: The Pivnichnyi (Northern) Bridge has been hit, following repeat strikes on another key Kyiv crossing. Russia is systematically targeting urban infrastructure as winter approaches. BBC World
- Newcastle grand final crowd horror: A P-plate driver ploughed into Knights fans at a grand final farewell event, leaving 10 injured including a five-year-old girl in critical condition. The driver was arrested at the scene. Guardian AU
- RBA rate hike bites — and Varoufakis is furious: With the RBA hiking again, former Greek finance minister Yanis Varoufakis has a characteristically blunt take: the rate rise is "the wrong medicine and an even worse alibi," accusing it of being used to vandalise Australian institutions. Whether you agree or not, it's worth reading. Guardian AU
- AUKUS submarine deal under scrutiny: Pete Hegseth called Virginia-class subs the "apex predator" of the seas, but Australian experts are pushing back — arguing these vessels are optimised for US-style force projection, not defending Australia's own backyard. Guardian AU
- Apple clamps down on AI agent overreach: Apple has tightened full-disk access permissions specifically to curb AI agents — including Meta's — from reading messages and sensitive data. A quiet but significant privacy move in the agentic AI era. Ars Technica
Markets
It's a tale of two markets today. The ASX copped a 3.3% shellacking — one of its uglier single-day drops in recent memory — while the AUD slid to 69.6 US cents, off 2.66%, suggesting macro risk-off pressure is hitting Australia harder than offshore peers. Meanwhile, the S&P 500 nudged up 0.73% and the Nasdaq surged 3.71%, likely driven by a tech relief rally, possibly on rate expectations or earnings optimism. Nikkei also ripped 6.19% — a monster session. Gold fell sharply (nearly 6%), which is unusual in a risk-off environment and worth watching. Crypto is catching a bid: Bitcoin up 4.3% to ~$84.7K and Ethereum jumping 7.1% — possibly rotating flows from gold or a short squeeze.
Worth a Read
- ADHD, Autism or Complex Trauma? — A Cambridge academic paper doing the rounds on Hacker News with 173 upvotes. The diagnostic overlap between these three is genuinely complicated, and misdiagnosis has real consequences. Worth a read if you or anyone you know has been caught in this maze.
- Our AI Midwife — Scott Alexander's latest on AI in healthcare. As agentic AI starts making real-world decisions, this piece asks hard questions about accountability and trust. Thought-provoking for anyone in tech.
- Anthropic lobbied the Pope to argue AI could be a conscious being — Yes, really. The Telegraph is reporting Anthropic made its case to the Vatican. The HN comments are predictably spicy. Wild sign of the times.
- Roosters and the Rabbi — five days before the Bondi attack — A quietly moving piece from ABC News about a Roosters visit to a Bondi synagogue just before the attack that shocked Australia. Worth a few minutes of your time.