US - China AI gap about to vanish

Latest Chinese frontier AI models have moved into the top tier of global rankings, with leading US and Chinese systems now separated by only a small margin. While the US frontier models remained clearly ahead in 2023 and 2024, the ones from China have risen much faster and are approaching the same level. Fully closing the gap would have major consequences, including for the very significant public and private investments currently made into any one of these models.

AI must now identify itself

From 2 August 2026, the EU AI Act makes AI transparency an enforceable obligation. Chatbots must disclose when users are dealing with AI systems – and AI content needs to be labelled as such, in a way that humans and machines alike can take note. What looks like a disclosure rule may prove something more fundamental: an early attempt to build a provenance layer for synthetic content. Whether such durable transparency about AI authorship is attainable needs to be seen. Adoption rate, interoperability and resistance to falsification will all play a role.

AI agents in global supply chains

A new World Economic Forum policy paper addresses how to control decision making in global supply chains in times of autonomous AI agents. An autonomous AI agent might reroute a shipment, switch suppliers, reallocate inventory or commit expenditure - potentially across companies and jurisdictions, and without a human approving every step. This makes the necessary governance less about controlling the AI model and more about defining which party may delegate which decisions, within what limits, and who remains accountable when an agent gets it wrong. Providing a policy framework for these questions is particularly difficult as the current distribution of roles among different parties of a complex supply chain may not subsist in their current form – the architecture of the supply chain itself will be changed.

Consumer protection in times of AI shopping agents

Most consumers are already using AI  to search for products and soon AI agents will buy items too. So the law must become "agent-ready".

This summer, U.S. Senator Mark Warner released a discussion draft of the AI AGENT Act. It would let consumers authorize a "custodial user agent" on a documented, scope-limited, and revocable mandate. It would also oblige large platforms to open their interfaces to such agents on fair, reasonable, and non-discriminatory terms, and impose non-waivable duties of loyalty and care on providers. However, what’s left  open is when an agent's act binds the consumer, and who bears the loss when it oversteps.

The EU's Digital Fairness Act proposal, which is due later this year, targets dark patterns, addictive design and unfair personalization. That is exactly where consumer agents operate. A loyal one protects the consumer. A conflicted one buys before they grasp the choice.

So far, both sides of the Atlantic land in the same place: providers owe duties, consumers keep control, independent agents get access. For anyone selling online, the rules now taking shape decide who may speak to your customers, and who is bound when an agent hits "buy". We will keep following this debate. It will gain traction quickly, because what is being settled here is a multi-billion-dollar question.

16 August 2016 – China launches world’s first quantum satellite

10 years ago, on 16 August 2016, China launched Micius, the world's first quantum-communications satellite. Named after the ancient Chinese philosopher Mozi who conducted optical experiments some 2,400 years ago, this satellite tested, for the first time ever, quantum key distribution, particle entanglement and quantum teleportation over large distances in space – something impossible via fibre cables on the ground. What began as an audacious physics experiment also foreshadowed today's contest over quantum-secure infrastructure – something that is now seen by most nations as a matter of utmost relevance for their digital sovereignty.