The most important news pieces from the AI world are here for you, whether you’re an AI enthusiast or just wish to keep an eye on what’s going on.
Focus Topic: The shutdown the world is still talking about
Last week I wrote about the US ordering Anthropic to block all non-US access to its two most capable models, Fable 5 and Mythos 5, citing national security and a jailbreak the company called minor. Normally a story like that fades in a few days. This one did the opposite: a week on, it was still the conversation – in the security community, at the G7, and across Europe’s debate over who controls the AI it runs on. So this is less a new story than proof of how big the first one was.
The loudest reaction came from the people who defend systems for a living. More than 100 cybersecurity executives and researchers signed an open letter (freefable.org) urging the US to lift the ban, arguing it handcuffs defenders without slowing attackers, who get the same capabilities from rival models. Alex Stamos, the former Facebook security chief, noted the flagged jailbreak had actually produced a tool defensive teams use to patch weak spots – and that GPT-5.5, Kimi 2.7, Opus and Sonnet all share the same capability. Their ask: regulation grounded in scientific evaluation and transparent enforcement, not a sudden shutdown.
Then the same names ended up in the same room. At the G7 summit in France, the heads of the leading labs – Sam Altman, Dario Amodei and Demis Hassabis – sat alongside heads of state, the most senior AI gathering in a long while. Trump said he no longer sees Anthropic as a national security threat, two days after meeting Amodei, yet the two sides remained at an impasse over the export limits, with the US reportedly refusing to exempt even G7 allies.
What the labs proposed there matters more than the standoff. Amodei and Hassabis pushed for a US-led coalition to set global AI rules, covering structured access to frontier models and chip trade that excludes China, while Altman called for an international forum on testing standards. In other words, the companies being regulated were the ones drafting the rulebook – and proposing who gets in and who is locked out.
For finance and business leaders, the takeaway has only hardened. The AI in your stack is now a geopolitical dependency, not just a vendor choice. A model your team relies on can be switched off by an export order with no notice, and access can hinge on where you operate or which passport your staff hold. The question is not model quality. It is continuity, concentration and fallback: what breaks if your primary model goes dark, and what is your backup.
LLMs & AI Models
- Z AI released GLM-5.2, an open-weights model matching GPT-5.5 and Claude Opus 4.8 on coding benchmarks – the closest open model yet to the top closed systems.
- Alibaba unveiled AI models for robots, as China’s focus shifts toward agents.
- OpenAI is preparing a major ChatGPT voice upgrade built on a new model called GPT-Bidi-1.
- Microsoft launched Copilot Cowork, a task-running AI agent, worldwide with usage-based pricing.
- Anthropic released Claude Fable 5, opening its top Mythos tier to the public with state-of-the-art benchmark scores, free across Claude tiers until June 22 before moving to usage credits. Days later it paused access to both Fable 5 and Mythos 5 after US government security concerns
- Google released DiffusionGemma, an open model that writes text in parallel chunks, quadrupling speed and hitting 1,000+ tokens/second on a single Nvidia H100.
- ChatGPT added scheduled-task controls, edging further toward becoming your AI personal assistant.
New Tools
- Facebook is rolling out AI Mode, letting Meta AI answer from public posts, Groups and Reels instead of a list of results.
- Midjourney announced the Midjourney Scanner, a machine that scans your whole body in 60 seconds using ultrasonic waves.
- Anthropic shipped a major Claude Design overhaul with design-system imports, code round-trips and a fix for its token-burning.
- Cursor introduced Origin, its own GitHub competitor, plus a new iOS app in beta for its AI coding platform.
- OpenRouter launched Fusion, an API that pools several models into one answer, nearly matching Fable 5’s deep research at lower cost.
- Android 17 shipped with new multitasking tools as Google expands Gemini across the platform.
- Epic Games took a big step toward AI-built games with Unreal Engine 5.8.
- ByteDance released Seedance 2.0 mini, a cheaper variant of its powerful AI video model.
- Adobe gave Firefly AI Assistant new agentic skills, now in beta across Photoshop, Premiere and more.
- Sakana AI launched Marlin, an autonomous research agent that can work up to 8 hours in one run.
👉 Explore these tools: Cursor | OpenRouter Fusion | Adobe Firefly
Other Quick Picks
- SpaceX is buying Cursor-maker Anysphere for $60B in stock, and Elon Musk became the world’s first trillionaire after its IPO.
- OpenAI’s net loss jumped nearly 8x to $38.5B in 2025, on $13B revenue against $34B in costs.
- DeepSeek raised $7.4B at a $50B+ valuation, becoming China’s priciest AI startup.
- Microsoft may put China’s DeepSeek inside Copilot to tame its AI bills.
- Anthropic faces a federal lawsuit alleging it oversold paid Claude subscriptions versus actual usage.
- Salesforce acquired Fin (formerly Intercom) for $3.6B, adding support agents and 30K customers to Agentforce.
- Apple’s Tim Cook said price hikes are unavoidable as AI demand drives up memory and storage chip costs.
- ChatGPT’s market share slipped below 50% for the first time.
- Nvidia is raising at least $20B in bonds, its first sale since 2021, to fund the AI buildout.
- US regulators ordered grid operators to fast-track AI data centers while keeping costs down.
- Norway is barring AI for children aged 6-13 in elementary schools from late August.
- The UK is set to ban social media for under-16s, following Australia’s lead.
- Meta began unwinding its $2B Manus deal after Beijing ordered the acquisition reversed.
- Anthropic’s study of 400K Claude Code sessions found domain expertise matters more than coding skill.
🇪🇪 AI News from Estonia
- Portugal’s defence-tech firm Tekever opened a Tallinn office and began work with Skeleton Technologies on defence, aviation and space tech.
- KPMG Baltics won the €6.45M tender for Estonia’s new state-budget information system, built on the SAP platform.
- MarkeDroid and STACC launched a microgrid project to let solar producers sell power to the grid more flexibly.
- A three-day Estonia-Ukraine GovTech hackathon in Kyiv gathered 80+ AI experts to prototype public-sector solutions.
- Estonia is building a digital “AI ID” for AI agents, so they can act on someone’s behalf within clearly defined, auditable rights.
- TalTech’s Europe Legal Design Action won EU funding to 2030 to build a 38-country legal-design network.
- Estonian experts read the US ban on Anthropic’s models for foreigners as a sign of tech decoupling, but expect it resolved within weeks.
- TalTech and Telia signed a partnership centred on AI, cybersecurity, quantum tech and defence.
🎙️ Inside Finance Podcast dropped a second episode with Sylvain OULALA, an experienced street artist and entrepreneur, on how to actually grab and hold attention in a distracted world and why social skills are becoming the rarest competitive advantage in business.
🎧 Listen to Inside Finance Podcast on YouTube and Spotify
🎧 Listen to the podcast on Apple Podcasts.
Want to stay in the loop?
Straight to your inbox: practical AI updates, finance use cases, tools to try, and upcoming trainings.
I cut the noise and send only what’s actually useful.
💼 Exploring how to make AI work in finance? Let’s talk practical use cases – connect with Gerlyn Tiigemäe for expert guidance.
📚 If you would like to participate in one of my trainings or listen to speaking engagements, here are the upcoming ones:
- Training “AI võimalused finantsvaldkonnas” @ Äripäeva Akadeemia, next trainings for beginners @ 31.07 – information, for intermediate level @ 21.08 – information
- Training “AI võimalused projektijuhtidele” @ Äripäeva Akadeemia, next trainings @ 29.06 – information
- I will be doing couple of training together with Eesti.ai algatus , which are all free, so check out their website to find suitable trainings here.
