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: Who actually owns the AI you build on?
This week the question of who controls AI stopped being a cap-table detail and became a public one. Three stories made it concrete.
First, the US government was reported to be discussing an ownership stake in OpenAI – the state potentially becoming a shareholder in the company that builds the models millions of businesses now run on. That is a different kind of oversight than regulation: an owner sits at the table.
Second, OpenAI published “Built to benefit everyone,” a blog from Sam Altman and chief scientist Jakub Pachocki that reframes the company’s mission in public-good language, arriving the same week it filed confidentially for an IPO. The timing is worth noticing: a company preparing to sell shares to investors is also telling the public it belongs to everyone.
Third, Argentina drafted a law for a “non-human corporation,” a company that would be owned and operated by AI itself rather than by people. Three countries, three motives, one shared question: who gets to own and steer these systems, and on whose terms.
Anthropic made the same point by accident. It opened its top Mythos-class model to the public as Claude Fable 5, the most capable model it has ever shipped. Days later it paused access to both Fable 5 and Mythos 5 after US government security concerns, and it had already apologized for safety filters that quietly downgraded some answers for researchers. So within a single week the best model on the market could be opened, throttled for safety, and then pulled – partly by the vendor, partly by a government. Nobody asked the businesses already building on it.
That is the real lesson for a company. If your workflow runs on one closed frontier model, access is a dependency you do not control. A price change, a new safety filter, an export rule, or a government pause can all reach straight into your process, and none of them are on your roadmap. The honest question is whether you have a Plan B. This is why public and open-weight models are getting more attention than the benchmark race suggests – not because they score higher, but because nobody can switch them off for you, and you can keep a copy running on your own terms.
For finance leaders this is supplier and continuity risk, not a tech-team footnote. The model behind an invoice-processing flow or a customer-facing assistant is now a critical vendor. Treat it like one: a second source, a documented exit plan, and a clear-eyed view of who actually controls the thing your numbers depend on. Ownership of AI is no longer an abstract debate. It shows up the day your access changes and you find out how much of your operation you had quietly handed over.
LLMs & AI Models
- 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.
- OpenAI is planning its biggest ChatGPT revamp yet, rebuilding it into an agent-and-coding “superapp” centered on Codex, with one insider reportedly saying “Chat is dead.”
- Visa is wiring its payment network into OpenAI’s agents, letting them spend within user-set caps and merchant rules while Visa handles fraud, chargebacks and refunds.
- Anthropic apologized after Fable’s safety filters silently downgraded answers and blocked benign queries; it keeps the guardrails but will now make refusals and reroutes visible.
New Tools
- Apple rebranded its assistant as Siri AI at WWDC 2026, built on Apple’s own models with Google’s Gemini, able to reason from on-screen content and take systemwide actions.
- Coinbase launched an AI trading agent that executes crypto trades and pays for research data, running inside ChatGPT and Claude through its MCP server.
- Google gave NotebookLM agentic chat, handing each notebook a sandboxed computer to write and run code and output PDFs, spreadsheets and slides.
- Google launched Gemini 3.5 Live Translate, a real-time voice model covering 70+ languages while preserving a speaker’s tone and pacing.
👉 Explore these tools: Siri AI | Coinbase AI agent | NotebookLM | Gemini 3.5 Live Translate
Other Quick Picks
- OpenAI confidentially filed a draft S-1 (IPO plan), a week after Anthropic’s, with Altman reportedly telling staff it plans to go public “within the next year.”
- Dario Amodei’s essay “Policy on the AI Exponential” urges faster regulation, powers to ground frontier models, and a jobs plan for mass automation.
- Anthropic’s CFO says his team built 150 Claude skills and gets the monthly financial review 90-95% done before a human edits it.
- SpaceX previewed AI1, a solar-powered satellite running AI chips in orbit, with Google and Anthropic already signed on as compute customers.
- SpaceX’s IPO drew nearly 4x demand, over $250B in orders against a $75B target at a $1.8T valuation; Senator Warren asked the SEC to delay it. Read more about it in my post.
- China is drafting a $295B five-year plan for a national AI data-center grid, sourcing 80%+ of core tech locally to sideline US chipmakers.
- China is pushing foreign AI models out over security concerns, even as US firms keep adopting cheaper Chinese ones like DeepSeek.
- Jeff Bezos raised $12B at a $41B valuation for Prometheus, an “artificial general engineer” meant to speed designing physical machines 10x.
- Nearly a third of firms cut a role for AI then rehired the same job – an “AI boomerang” led by finance at 44%.
- The EU ordered Meta to restore free WhatsApp access for rival AI assistants during its antitrust probe; Meta calls it overreach and will appeal.
- The UK launched a £1.1B AI Hardware Plan to back homegrown chip startups, including £750M for a national supercomputer.
- Argentina proposed a “non-human corporation” owned and run by AI, with liability protection and tax perks; Harari warned of an “AI state.”
- OpenAI’s “Built to benefit everyone” sets out a “third phase,” a personal AGI for all, and a global body that could pause frontier AI work.
- A Perplexity-Harvard study found AI agents pull users toward more ambitious work, with 59% of tasks falling outside the user’s own field.
- New York became the first US state to require ads to disclose AI-generated actors, with $1,000 fines and SAG-AFTRA backing.
- Lionsgate is taking a stake in AI video company Runway to co-develop IP and short-form projects from its catalog.
- The 2026 FIFA World Cup opened with AI across offside calls, analytics and fan tools, capturing 150M+ data points per match.
🇪🇪 AI News from Estonia
- The creator of Tekstiks.ee built a new speech-tech app: jutusta.ee, entirely self-made and self-hosted, and plans to turn it into a research-based company.
- Bolt began a self-driving car pilot, treating the city as a “living lab,” with a goal of bringing 100,000 cars to market.
- The 28-company Hansa Grupp is investing ~€1.6M in IT this year on data platforms and AI, though its CEO expects no real net savings yet.
- Iizi launched a motor-insurance price calculator inside ChatGPT, which it calls Estonia’s first insurance tool in OpenAI’s app ecosystem.
- The Institute of the Estonian Language and the Estonian Actors’ Union signed a deal requiring consent before commercial use of actors’ synthetic voices.
🎙️ 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
- 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.
