White House AI Safety Tests: Labs, Cyber Rules & TSMC

The State of Tech, The European Edition of Tuesday 4 August 2026


White House convenes OpenAI, Anthropic, Google and Meta to build voluntary AI cybersecurity safety tests

CrowdStrike 2026 Threat Report: AI is now weapon, target and force multiplier in modern cyberattacks

TSMC targets fivefold jump in two-nanometre chip output by end of 2026 as AI demand surges

Stanford researchers urge urgent regulation of world models before they embed in cars and infrastructure

LibreLingo: the free open-source language learning app that works offline and needs no account

Turtl: the Danish encrypted note app that keeps your data private across every device