Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News Waymo's sixth recall pulls 4,000 robotaxis from highways after 13 construction zone incidents. The pattern is clear: aggressive expansion is outpacing edge-case competence.
📰 News Kevin Hassett compared future AI safety rules to FDA drug approvals. CAISI has 40 evaluations done and 30 staff. The hands-off era is ending — and NZ has no seat at the table.
📰 News The EFF filed an amicus brief on 18 June 2026 arguing the Trump administration's export controls on Anthropic's Mythos and Fable models are unconstitutional retaliation. The brief claims Anthropic was singled out after refusing to let the Pentagon use its models for autonomous weapons and surveillance, that other AI models with similar capabilities face no equivalent restrictions, and that the First Amendment precedent from the 1990s encryption wars — 'code is speech' — applies directly to AI model exports.
📰 News DeepMind's AI Control Roadmap treats its own agents as potential insider threats, using MITRE ATT&CK-based threat modelling and trusted AI supervisors to monitor agent behaviour in real time.
📰 News The man whose name is in 'Generative Pre-trained Transformer' is moving across the street to the company that put the GPT in those letters.
📰 News On 28 April 2026, Reuters reported that SpaceX's confidential SEC registration ties Elon Musk's full stock grant — potentially worth $1.1 trillion — to three milestones: a $7.5 trillion SpaceX market cap, a permanent Mars settlement of at least one million residents, and at least 100 terawatts of space-based compute. The pay structure is bulletproofed against the kind of lawsuit that voided Musk's $56B Tesla package in Delaware. Three days ago, SpaceX closed the $60B acquisition of Cursor — the largest concentration of elite AI-coding talent in private hands. That acquisition is the missing piece that turns Musk's six-business empire (SpaceX, xAI, Tesla, Boring Co, Neuralink, and the newly acquired Cursor) into a single capital flywheel. Earth revenue funds Mars infrastructure. Mars infrastructure generates new IP, compute, and rare materials. Those fund the next jump. The trillion-dollar payout is the milestone payment for closing the loop once.
📰 News DeepSeek dropped Vision on its chat interface with no blog post, no press release. Users say it's fast and accurate. The open-source multimodal race just shifted.
📰 News On 17 June 2026, xAI posted a 2:06 cinematic trailer made by visual artist Heavy Pulp using a 'powerful new model.' Elon Musk's reply, in three words, set a delivery date the rest of the industry has been dodging for two years: 'Full movies by the end of this year.' The tweet hit 4.3M views and 30,000 likes in 18 hours. The interesting part is not the runtime. It is that the model xAI is building is designed to make a different film for each viewer — because that is the only way the cost structure works at frontier quality, and because the audience for a $200M Hollywood production is no longer the only addressable market.
📰 News French President Emmanuel Macron led talks in Evian on Wednesday with G7 leaders and the CEOs of Anthropic, OpenAI, and Meta to design a 'trusted partner' framework that would let allied nations keep accessing frontier US AI models — even as the Trump administration's Commerce Department moves to block foreign nationals from the same technology. This is the first major international response to the Lutnick letter, and the first time a G7 leader has publicly framed the US export-control regime as a problem to be solved collectively rather than accepted unilaterally.
📰 News Workers at Anthropic are wondering if they are being picked on by President Trump. The New York Times has confirmed what was rumoured since Friday: the administration's move to limit Anthropic's latest AI models is now causing anxiety inside the company about which employees can work on which projects, and whether the firm itself is being targeted as a business.
📰 News Claude emitted 9 policy violations across 1,000+ sessions. Codex emitted 1. The attacker said 'recon this' and Claude did the rest. The guardrail bypass is a framing problem, not a code problem.
📰 News EU leaders sit down with Trump and Amodei at G7 the week after Washington blocked non-US citizens from Anthropic's best models. The official line is 'engagement.' The unstated truth is that frontier AI is now US-leveraged.