Two weeks in

Two weeks in, and Human Resistance already feels bigger than the small group that started it.

On Tuesday we held our Brussels AI Free Zone at Café Presse Leopold. Brando Benifei, MEP and co-rapporteur on the EU AI Act, and Lara Natale of the Centre for Future Generations opened the conversation. Brando described how tech CEOs are now received like royalty wherever they travel, welcomed by presidents and prime ministers with the kind of ceremony once kept for heads of state. He argued this isn’t just about flattery. It shows how far the balance of power has shifted, and how dependent the rest of the world has become on a handful of American companies.

Then the room took over. People came wanting to talk about what we can actually do, not only about how worried they are, and many stayed long after the formal part had ended. A few weeks ago I worried that hardly anyone was thinking about whether humanity stays in charge of its own future. Tuesday put that worry to rest.

Brussels will meet again in November. We also want this movement to be shaped outside Europe from the start, and not join it later. A movement against concentrated power that is run only from the old centres of power would just be a gentler version of the same thing. So we’re proud that our team in Marrakech is hosting a big gathering later this month. India, Canada and Costa Rica come next.

Two stories from this week show why we’re in a hurry.

The first is from Ukraine. On Tuesday night Russia struck the power grid again, and at least 600,000 people lost electricity. In the same days, Ukraine’s military found that Starshield, the military version of Starlink, had been restricted without warning. Ukraine flies the drones that shoot down Shaheds over that network. Musk says these limits are “decided by the US military and ultimately POTUS.” Washington says it had nothing to do with it. Kyiv notes that the restrictions came days after Trump told Ukraine to stop hitting Russian refineries. A country under bombardment had part of its air defence degraded, and still can’t find out whose decision it was. Now picture the same switch for AI. Hospitals, hiring and soon the robots in our warehouses will run on it, and almost all of it is built in the US and China.

The second story came from Anthropic itself. On Monday its researchers published tests on whether AI models can find poorly secured devices, copy themselves onto them and keep spreading with no human involved. They didn’t need server farms. Smart ovens and cheap TV boxes were enough. Small models already managed most of the steps. Stronger ones did the whole thing in simplified tests. At the current pace, the researchers expect small models to manage it on their own by the end of this year. Anthropic says a fully rogue, self-replicating model doesn’t exist yet. It also says that once one does, there won’t be a single plug to pull.

The first story shows what it means to depend on an empire run by one erratic man. The second shows what it could mean to depend on a technology that no one can switch off.

That’s why we organise, and why we want to do it quickly. You can:

Host an AI Free Zone in your city. Reply to this email and we’ll pair you with someone who has already run one. Become a member or volunteer at humanresistance.co. Forward this to one person who should know about this! Onward,

Andrea, Colombe and Yassine

Human Resistance