Thursday, June 18 June 18, 2026
G7 leaders clash with the U.S. over AI kill-switch risks as Anthropic's model export ban sparks a sovereignty crisis; a new Pew study finds only 16% of Americans trust AI's impact; world model startup Odyssey hits $1.45B valuation backed by Amazon; and Noam Shazeer, co-lead of Google Gemini, defects to OpenAI.
Good morning. It's Thursday, June 18th, 2026, and today's theme is trust — or the lack of it. Governments don't trust the U.S. to keep AI models accessible. Americans don't trust AI to benefit society. And enterprises still can't trust AI enough to move it out of the pilot phase. Let's get into it.
We start with the biggest geopolitical story in AI right now. At the G7 summit yesterday, world leaders including France's Emmanuel Macron and India's Narendra Modi confronted U.S. officials — including President Trump, Sam Altman, and Dario Amodei — over AI access. Macron warned that if the U.S. can "turn off the switch" on American AI models overnight, it would harm European economies and ultimately damage the AI companies themselves. This comes after the Trump administration blocked Anthropic from exporting its newest Mythos 5 and Fable 5 models on national security grounds. G7 leaders are now floating a "trusted partners" scheme to guarantee allied nations continued access. The message is clear: we want your AI, but we need a kill-switch-free version of it.
Next, a striking new Pew Research study. Only 16 percent of Americans believe AI will have a positive impact on society over the next 20 years. About 40 percent say the impact will be negative. Nearly two-thirds think AI is moving too fast. And here's the irony: about a quarter of Americans use AI chatbots every single day. People are using it — they just don't trust it. ChatGPT leads adoption with 44 percent of U.S. adults, followed by Gemini at 24 percent, Copilot at 17 percent, and Meta AI at 14 percent.
On the funding front, world model startup Odyssey just hit unicorn status — a 310-million-dollar Series B at a 1.45 billion dollar valuation. Amazon, AMD Ventures, and GV all participated. Odyssey's founders come from self-driving backgrounds — the CEO previously led Cruise's product team at GM. Their approach: send people out with cameras to capture the physical world, then simulate it with accurate physics for use cases from robotics to video game creation. With Amazon's backing, AWS becomes Odyssey's preferred cloud and they'll optimize for Trainium chips — a direct challenge to Nvidia's grip on AI infrastructure.
In AI reliability news, Pramaana Labs raised a 27-million-dollar seed round led by Khosla Ventures to bring formal verification to AI. The startup targets high-stakes verticals — law, drug discovery, tax — where hallucinations aren't just annoying, they're costly. Their approach layers a deterministic verification system on top of a standard LLM using the LEAN programming language, the same tool used to verify mathematical proofs. Former IRS Commissioner Danny Werfel is already advising on the tax law module. Enterprises struggling to move AI out of pilots will be watching this closely.
Big talent move: Noam Shazeer — co-lead of Google's Gemini project — has joined OpenAI. Shazeer spent two decades at Google, left to co-found Character AI in 2021, then returned to Google in 2024 after they reportedly paid 2.7 billion dollars to bring him back. Now he's at OpenAI. When one of Gemini's chief architects crosses the aisle, the model race gets even more interesting.
Adobe made a wave of announcements this morning. Photoshop and Premiere are both getting prompt-based AI assistants. The redesigned Firefly AI studio now remembers your past creations for brand consistency. And Adobe is collaborating with Disney Imagineering to bring Firefly into theme park design workflows. Adobe is making a hard push to be the AI creative layer for professional teams.
Finally, Anthropic quietly became the first AI startup to join the Frontier carbon removal coalition. It's a signal that at least one major AI lab is trying to get ahead of the energy narrative — especially as data centers face growing community pushback across the U.S.
Here's your one business idea for today. The Pew data — 84 percent of Americans skeptical of AI — creates a real opening for an "AI Trust Audit" consultancy. Target mid-market companies deploying AI in customer-facing roles: healthcare portals, financial advisors, HR systems. Offer structured third-party reviews certifying deployments as transparent, accurate, and explainable. With regulators circling and public trust at historic lows, a credible "this AI is trustworthy" stamp is something companies will pay for.
That's your briefing for June 18th. The AI industry is moving fast — but the world is asking it to slow down and earn some trust first. Have a great day.