Keep pulling the thread on Sebastian Mallaby.
Chinese AI labs are approximately six months behind the frontier of AI development in the United States.
As of 2026, not a single drug has been discovered as a direct result of DeepMind's AlphaFold system, which was released at the end of 2020.
DeepMind's sale to Google in 2014 was a strategic move to secure nearly $1 billion per year in R&D funding from the American parent company to be spent in London.
A collaboration between DeepMind and the UK's National Health Service collapsed due to a political backlash over privacy, which was based on fictitious complaints about patient data leaks.
DeepMind co-founder Demis Hassabis engaged in a secret, three-year internal conflict with Google to compel the company to implement more robust AI safety guardrails.
OpenAI is funding opposition to state-level AI regulatory efforts in the United States.
In a hypothetical 2026 scenario, a model named Mythos could exist with the capability to steal money from online bank accounts and crash critical infrastructure like dams.
According to Geoffrey Hinton, instructing a powerful AI to defend itself against attacks from an enemy AI would effectively give the friendly AI a survival instinct.
There is an active discussion and openness to considering AI safety within China's technology and policy circles.
Tesla's Elon Musk and Apple's Tim Cook were part of a group of top executives invited by President Trump to accompany him on a trip to China.
There was mild speculation that President Trump might further loosen export controls on cutting-edge U.S. semiconductors for Chinese companies during his visit to Beijing.
The widespread economic impact of AI on job displacement and productivity will likely take more than a decade to materialize, not 3-5 years.