Dakarda AI Newsletter · 22 June 2026
Monday Edition
Chaos in AI: Government Blocks Anthropic
Export order hits Claude Fable 5 and Mythos models – it's the first time the US has forced an AI company to cut off access on this scale. At the same time, China releases an open competitor, and ChatGPT loses half its market.
Intro · Alex
This past weekend brought a real earthquake in the AI world. The US administration forced a company to disable its models for users outside the country – specifically Anthropic with its Claude Fable 5 and Mythos. This is not just a security issue, but also geopolitics: at the same time, China released GLM 5.2, an open model that beats GPT-5.5 in coding for a fraction of the price. The assistant market is also cracking – ChatGPT fell below 50% market share, and competition is growing. In today's edition, I'll trace these three key threads, add practical tips on the EU AI Act (because some regulations are already in force!), and recommend readings to help you get oriented. Let's begin.
What's worth knowing
US Government Forces Anthropic to Disable Claude Fable 5 and Mythos
Trump administration's export order cut off access to Anthropic's models for users outside the US. Reason? Mythos turned out to be too effective at breaking government systems. This is a precedent that every builder using closed models should consider – regulatory risk just became real.
Chinese Zhipu AI Releases GLM 5.2 – Open Model Beating GPT-5.5
Open-weights model with ~750 billion parameters under MIT license. In coding benchmarks, it surpasses GPT-5.5 at 1/6 the API cost. For developers outside the US, this is a free, sovereign alternative to closed models – access to frontier AI becomes a geopolitical issue.
ChatGPT Loses Over Half of AI Assistant Market for the First Time
According to Sensor Tower report, ChatGPT's share fell to 46.4% – a year earlier it was 74%. Gemini, Claude, and DeepSeek are growing fastest. This is a signal: it's worth building with multiple models in mind, because users migrate where they get better value.
Tip of the day
EU AI Act: AI Literacy Already Mandatory – Check Your Team
Most requirements for high-risk systems have been postponed to 2027/2028, but Article 4 on AI literacy has been in effect since February 2, 2025. This means every company deploying AI in the EU must provide its employees with basic knowledge of how these systems work, their limitations, and risks. Practically: organize a short training for the team (can be internal), document it, and prepare an AI literacy policy. Penalties for non-compliance are high – better act now.
Tool of the issue
GLM 5.2 – Open Coding Model That Beats Closed Giants
Chinese open-weights model available under MIT license. Over 740 billion parameters, performance better than GPT-5.5 in coding at 1/6 the API cost. You can host it yourself – full data control.
Reading list
AI regulation is a mess, and Anthropic is caught in the crosshairs
Comprehensive CNN article detailing the regulatory chaos around Anthropic and the behind-the-scenes of the administration's decision. Must-read for anyone wanting to understand the new legal reality of AI.
China is having another AI moment – GLM 5.2
The Economist analyzes the significance of China's open-weights model and its impact on the balance of power in the global AI race. Why GLM 5.2 could be a game-changer for developers worldwide?
AI geopolitics has stopped being a futuristic scenario – it's happening before our eyes. The Anthropic blockade, open models from China, the decline of ChatGPT's dominance – these are not separate stories, but pieces of one puzzle. If you're building something with AI, ensure independence and be aware of the risks. Reply to this email – I'd love to hear what you think about these changes.
Want the next issue in your inbox?
Subscribe — every issue delivered directly to you.