This Week’s News

  1. Anthropic faces emergency action by the U.S. government and scrutiny at home:

    1. The Trump administration issues an emergency ban blocking all foreign nationals from accessing Anthropic’s most powerful AI models, effectively pulling them globally. [Link]

    2. Anthropic calls for a global voluntary pause on frontier AI development, warning that recursive self-improvement could strip humans of meaningful control. [Link]

    3. After backlash over a quiet downgrade to Fable 5 for undisclosed national-security reasons, Anthropic pledges to explicitly inform users when requests are rejected or rerouted. [Link]

  2. Apple’s WWDC 2026 reveals a rebuilt Siri powered by a custom Google Gemini model and a new AI Extensions framework letting users route iPhone tasks to ChatGPT, Claude, or Gemini with distinct voices. [Link]

  3. OpenAI confidentially files its S-1 for an IPO at an implied ~852 billion USD valuation, setting up a historic dual-AI-lab market debut race with Anthropic’s own filing earlier in June. [Link]

  4. Jeff Bezos’s startup Prometheus raises 12 billion USD to build an “Artificial General Engineer” for complex physical systems such as jet-engine design and drug-compound manufacturing. [Link]

  5. Meta abandons its open-source Llama strategy in favour of a new closed-source model, Muse Spark, developed by its newly formed Superintelligence Labs. [Link]

  6. German humanoid robotics firm Neura Robotics closes a 1.4 billion USD Series C backed by Amazon and NVIDIA, as humanoid robot makers collectively raise 55.8 billion in 2026. [Link]

  7. Chinese AI lab DeepSeek breaks its no-external-funding policy with a 7.4 billion USD maiden round, targeting a valuation of up to 59 billion USD. [Link]

  8. Moonshot AI releases Kimi K2.7-Code, an open-source trillion-parameter coding model that targets GPT-4.1 and Claude Opus on coding benchmarks. [Link]

  9. The EU AI Office releases its finalised Code of Practice on AI-generated content labelling under the AI Act, with full enforcement arriving on August 2, 2026. [Link]

  10. President Trump signs an executive order creating a voluntary framework for AI labs to engage the U.S. government pre-release, deliberately avoiding mandatory licensing requirements. [Link]

  11. Humanoid robots move from pilot to platform on real factory floors in 2026, with Figure 03 producing one per hour, Boston Dynamics’ Atlas shipping to Hyundai, and Amazon’s autonomous fleet crossing one million units. [Link]

  12. OpenAI rolls out GPT-5.5 personalisation to all free ChatGPT users, simultaneously retiring GPT-5.2 and migrating all legacy conversations. [Link]

  13. New York state sends seven AI-related bills covering transparency, discrimination, and consumer disclosure to Governor Hochul at the close of the 2026 legislative session. [Link]

  14. MiniMax releases M3, the first open-weight model to combine frontier coding, a 1M-token context window, and native multimodal video input. [Link]

  15. The EU’s Digital Markets Act blocks Apple’s Gemini-powered Siri from launching on EU iPhones, excluding around 450 million users from the new experience at launch. [Link]

  16. NVIDIA releases open-source GR00T N2 models enabling robots to follow natural-language instructions for complex multi-step physical tasks. [Link]

Ukrainian USV (Unmanned Surface Vessel)

Ukraine's Drone Playbook: Cheap, Clever, and Relentless

From garages and makerspaces scattered across the country, Ukraine has built a war-winning drone industry that's rewriting the rules of modern warfare. How has Kyiv turned plywood, code and crowdsourcing into weapons that impose devastating costs on Russia—and what does this distributed, software-first approach mean for the future of conflict?

Podcasts Your two favorite Deep Dive hosts discuss AI in depth, courtesy of NotebookLM