Newsletter

AI bosses crash the G7 🌍, Gemini drop nears 🤖, patch your PC 🔒

16 June 2026

Future Technology Daily Briefing

Future Technology

June 16, 2026 · Edition #4

Morning. The people who build the world's most powerful AI are sitting across the table from the people trying to regulate it, Google's big one is nearly here, and Goldman Sachs has put a number on the AI boom that will make your eyes water. Here's your three minutes.

Top Story

The AI bosses just walked into the G7

Source: The Next Web (3 minute read)

For the first time, the three people who actually build the world's most powerful AI are in the same room as the people trying to regulate it. The G7 summit opened in Evian-les-Bains, France, on Sunday and runs to Wednesday. Sam Altman of OpenAI, Demis Hassabis of Google DeepMind and Dario Amodei of Anthropic all turned up in person. Macron personally invited Altman. It's his first G7.

The trouble is the seven governments can't agree on what the rules should even be. The US doesn't want binding global AI agreements that might slow its lead. Europe is pushing its AI Act and wants hard rules. Canada's Mark Carney wants common safety standards everyone signs. Two years ago the G7 made firm commitments on AI risk. This year the language is expected to get quietly watered down.

The one concrete thing on the agenda is a working lunch with the lab bosses on Wednesday the 17th (around midday CEST / 11am BST). Everything else is photos and communiqués.

Why it matters: this is where we find out whether the richest democracies can agree on one rulebook for AI, or whether we end up with a patchwork of national rules that only companies big enough to hire armies of lawyers can navigate. Three rival labs lobbying governments face to face, at the same table, is brand new. Watch who they stand next to.

Quick Hits

Anthropic wants everyone to slow down. Its own AI didn't get the memo. The lab published a paper calling for a globally coordinated "pause" on frontier AI, while admitting that more than 80% of the code in its own products is now written by Claude, not humans. One of those facts rather undercuts the other. (SiliconAngle, 4 minute read)

Google's big one is nearly here. Gemini 3.5 Pro is expected to land before the end of June, with a 2 million token context window (roughly a whole bookshelf in one go) and a new "Deep Think" reasoning mode. Pichai said "next month" on stage on 19 May. Betting markets are eyeing the 23rd or 30th. (TechTimes, 3 minute read)

Update your Windows machine this week. Microsoft's June patch was one of its biggest ever, closing close to 200 security holes including several zero-day flaws that were already public before the fix shipped. If your PC keeps nagging you to restart, this is the week to let it. (The Hacker News, 2 minute read)

SpaceX is a stock now, and the index funds are circling. SPCX closed its first day at $168.70, up 25% on the IPO price, and passive funds are being forced to buy as it enters the major indices. Nasdaq-100 entry is about three weeks out. None of this is really about rockets. It's about index mechanics. (Investing.com, 2 minute read)

The number of the day: $7.6 trillion. That's Goldman Sachs's estimate for what the world will spend on AI between now and 2031, on chips, data centres and power. For scale, that's roughly a quarter of annual US GDP. Forward that one to whoever still calls AI a fad. (RD World, 2 minute read)

Tool of the Day

Kimi K2.7 Code

A free, open-weights coding model that landed last week and scores near the top of the tool-use benchmarks. The interesting bit isn't the score, it's that you can download it and run it on your own hardware, so no company or government can switch it off the way Anthropic's Fable 5 just got switched off. Honest version: skip this if you don't write code, and you'll want a serious GPU to self-host it properly. But if you build things and the past week made you nervous about renting your brain from someone else's cloud, it's worth a look. (Hugging Face, 3 minute read)

One genuine question before you go: if 80% of the code at a top AI lab is already written by its own AI, does a government "pause" actually mean anything any more? Hit reply and tell me. I read every one.

Nath,
Future Technology

Forwarded this? Get your own free briefing at futuretechnologyhq.com/newsletter.

Some links in this newsletter may be affiliate links. We only recommend products we genuinely think are worth your time.