In partnership with
News of the day
1. The EU launches a new action plan to regulate advanced AI, enhance cybersecurity resilience, and develop sovereign AI capabilities by 2027. → Read more
2. Allegations of China stealing AI capabilities highlight the need to distinguish authorized distillation from covert extraction, advocating for access layer security over blanket restrictions. → Read more
3. Microsoft's Copilot now favors GPT-5.6, yet strategically uses internal models for cost savings, highlighting a shift in AI resource management. → Read more
4. OpenAI CEO Sam Altman now believes AI creates more jobs than it eliminates, a shift from earlier predictions of mass layoffs. → Read more
Our take
Hi Dotikers!
Europe isn't waiting, it's building. Three weeks before the AI Act kicks in on August 2, the Commission is finally deploying a structured answer to accelerating autonomous threats. This cybersecurity and AI action plan is trying to turn dependency into controlled architecture.
The picture is unforgiving: frontier models now discover vulnerabilities and orchestrate attacks without a human check-in point. A machine finding zero-days, then exploiting them, then spreading. That's the scenario every critical operator loses sleep over. Rather than play prohibition or trust suppliers on their word, Brussels is imposing pre-market scrutiny: before a model touches European soil, it gets evaluated. An independent European capacity, operational by 2027, will handle it. Three million euros in audits is pricey, but nothing next to a European finance sector frozen by an autonomous attack.
What's striking in parallel is structured access. Ten days ago, Washington orchestrated a private preview of GPT-5.6, opaque processes, invisible criteria. Here, the EU enforces a transparent blueprint: who gets what, on what terms. Coinbase and Vercel already said it in July, you need a vendor mix, not single dependency. The EU is doing the same thing but at critical infrastructure scale. Banking, energy, health will get guaranteed and verifiable access to top-tier AI tomorrow, without routing through Washington's or Beijing's back door.
But there's a wry bit of irony: the Commission is launching this arsenal right as it's still building the castle walls the adversary already left. It's investing 300 million in AI Factories, training, defensive research. Admirable. Necessary. Yet the real frontier models stay, for now, owned by American and Asian players. Europe is negotiating access rather than building it. It could end up buying the same product, just with an "EU-approved" stamp. That's a start, not a win.
The timing is also tight. Three weeks to prep before AI Act enforcement hits every ministry, every bank with illegality risk. And when you see the implementation calendars (blueprint 2027, testing platform end of 2026, open-source campaign Q4 2026), you get it: Europe is chasing a ball it just dropped.
G.
Cut Lead Review From Hours To Minutes
Sign up for a free trial of Attio, the agentic CRM.
Ask Attio to build a daily workflow that surfaces the deals that need your attention today, like anything with a stage change, a recent reply, or a new signal in the last 24 hours.
Review your pipeline in Claude, synced live from Attio via MCP.
That's it.
Meme of the day
