Harald Klinke

Anthropic’s temporary shutdown of Fable 5 and Mythos 5 access for foreign users is a reminder that AI sovereignty is no longer just about chips and clouds. It’s increasingly about who controls access to frontier models. #AI #DigitalSovereignty #AISafety

Statement on the US government directive to suspend access to Fable 5 and Mythos 5 anthropic.com

3/3 The future question is not simply whether AI “replaces humans.” It is what happens when execution becomes cheap. Then the scarce resource is judgment: which problems matter, which results count, which systems may improve themselves, and which institutions can still keep up…

When AI builds itself anthropic.com

Websites are increasingly becoming: • machine-interpretable identity and knowledge interfaces • primary source layers for AI systems • infrastructures of trust and authority • strategic hubs serving both humans and agents. #AISovereignty #AIInfrastructure #DigitalTransformation

Google Search as you know it is over | TechCrunch techcrunch.com

AI is here to stay, but many LLM business models may not be. If massive investment fails to produce sustainable returns, the bubble—not the technology—could burst. #AI #LLMs #AIBubble

Ed Zitron: Apple Will 'Watch Everything Burn' When the AI Bubble Bursts macrumors.com

After scraping around 12 million artworks from Cara, the developer is now collaborating on Lantern, an open source tool helping artists track their work in public AI datasets. A fascinating turn in the debate over AI, provenance and control. #AIArt #Copyright #Provenance

He Scraped All of Their Art for AI. Now He’s Collaborating on a Tool to Help Them wired.com

How much of an AI coding agent’s performance comes from the model, and how much from its harness? A new study finds no clear overall solve rate advantage for vendor native harnesses, but substantial differences in cost, speed and tool use. #AIAgents #AICoding #LLM

Harness or Model? Isolating the Harness Effect in Agentic Coding with a Contamination-Controlled Private Suite arxiv.org

European AI sovereignty needs more than homegrown model. It also requires transparent training data, open weights, and independent infrastructure. Soofi S is an encouraging step in that direction. 🇪🇺 #AI #OpenSource #DigitalSovereignty

Soofi S — Sovereign German-English Foundation Model - a Hugging Face Space by Soofi-Project huggingface.co

Brazil is pairing US and Chinese AI compute with RISC-V and a national cloud. Sovereignty means diversifying dependencies. Europe should notice. #AI #Sovereignty

reuters.com

Inherent is training Faraday, a research agent based on a 27B Qwen model, to reproduce published research, choose experiments and use coding agents as tools. Its Replica benchmark results suggest that specialized agent training could matter more than model size. #AIAgents #AIR…

Training AI Scientists to Replicate Research inherentlabs.ai

Local first. Cloud when needed. Sync between. I believe this will become one of the defining AI architecture principles of the coming years. It combines privacy, resilience, low latency and scalable intelligence. #Qdrant #EdgeAI #AIInfrastructure #AISovereignt

1/3 Anthropic’s “When AI builds itself” describes a striking sequence: 2021–23: humans write code. 2023–25: chatbots suggest snippets. 2025–26: coding agents edit files. Today: agents run code and delegate work. Next? AI systems may help build their own successors.