Skip to main content

AI

Image
Ilustrační obrázek

Germany's Agile Robots raises $500M to scale humanoid production

Agile Robots closed a $500m Series F at an $8.5bn valuation on November 12, 2026. SoftBank led the round; Nvidia, Hitachi Group and T.Capital joined. The Munich company says the money will fund U.S. expansion and scale manufacturing in Germany, including a facility near Munich with capacity for 10,000 legged and wheeled humanoid robots a year.
November 12, 2026 Daniel Cesak
Image
Ilustrační obrázek

Uber booked four rides through AI agents despite massive weekly volume

Uber said it handled roughly 320 million trips and orders during Halloween week. Four of those rides went through an AI agent. Five hundred Uber Eats orders did too. That is the measured distance between the agent talk on travel earnings calls and what customers actually did with their phones.

Four rides out of 320 million

The numbers come from the companies themselves. Uber said on its Q3 earnings call that it handled roughly 320 million trips and orders during Halloween week, including four rides booked through AI agents and 500 Eats orders.

November 12, 2026 Daniel Cesak
Image
Ilustrační obrázek

First Time in 13 Years, G20 Talks Have No Seat for Civil Society

The Civil 20 was created in 2013 to give non-governmental organisations a formal voice in G20 talks. In 2026, for the first time since then, it does not exist. The European Commission says it has told the G20 presidency, in writing, that this is a loss.

For thirteen years, a seat in the G20 process was reserved for organisations that sell nothing: consumer groups, digital rights advocates, researchers, environmental NGOs. This year only one engagement group survived, and it represents business.

November 11, 2026 Miriam Česáková
Image
Ilustrační obrázek

Capital Markets Want AI Agents, but MiFID II and AI Act Set the Terms

A sponsored Harvard Business Review piece published on 10 November 2026, produced with AWS and Moody's, makes a narrow argument: capital markets have stopped asking whether AI agents work in demos and started asking how to keep them inside an audit trail. The HBR summary does not publish performance numbers. Its claim is narrower: Moody's positions institutional data and risk analytics as the layer that makes agent output checkable and auditable. For a European bank the notable part is not the demo.
November 11, 2026 Daniel Cesak
Image
Ilustrační obrázek

META-IR trains LLMs to audit their own reasoning for safety

A new arXiv preprint proposes META-IR, a safety framework that trains large language models to inspect their own intermediate reasoning for policy violations and adjust their risk checks when a jailbreak attempt changes shape. The paper is free to read. It is also not a compliance certificate for European deployers, who since 2 August 2026 face binding Article 50 transparency duties, including disclosure of AI-generated content and machine-readable marking.

META-IR stands for Meta-Introspective Reasoning for Adaptive Safety Alignment in LLM.

November 10, 2026 Daniel Cesak
Subscribe to AI
X

Don't miss out!

Subscribe for the latest news and updates.