This week two research repositories went public, each with a daily automated update.

AI Research tracks the AI revolution country by country: the US, China, the EU, Germany, France, the UK, Russia, Japan, Korea, India, the Gulf and the Philippines. Some of what stood out (sources in the repo):

  • OpenAI closed a $122 billion round at an $852 billion valuation (31 March 2026), and says it makes about $2 billion a month.
  • Big Tech’s 2026 data-centre spending is guided at roughly $635–745 billion, depending on the tracker.
  • About 97% of humanoid robots shipped in the first half of 2026 came from Chinese makers (one source; marked ⚠). Unitree’s Shanghai listing closed 460% up on its first day.
  • After a run of safety incidents in September, the heads of OpenAI and Anthropic both spoke in favour of slowing down; the White House said it would not halt AI development.

Cybersecurity Research is a study path from the fundamentals to AI security. Its first tool, the KEV watch, records each day which vulnerabilities CISA has confirmed are being exploited. On its first run the catalog held 1,756 entries. Five were AI infrastructure: three in LiteLLM (an AI gateway), one in MLflow and one in Ray. They were not exotic AI attacks. They were command injection, SQL injection, improper authentication, server-side request forgery and code injection, in software built quickly for the AI boom.

That’s the point of starting with fundamentals: securing AI is mostly securing the ordinary software around it, done carefully.