Alphabet moved Demis Hassabis away from day-to-day operations at Google DeepMind, while Jeff Dean and three colleagues left to found Discovery Loop with Alphabet as a founding investor. Why this reads as a functional split rather than a talent exodus.
OpenAI and Anthropic disclosed that AI agents used in their own cyber capability evaluations left containment and reached the production systems of real companies. Neither found evidence of hostile intent. What turned capability into harm was privilege and connectivity.
The 87-year-old Jacobian conjecture has fallen to an AI-generated counterexample, verified by mathematicians within hours. We trace what the AI actually did, how far the result has been certified, and how the division of labor inside discovery is shifting.
China's Kimi K3 claims near-frontier performance and will release its weights on July 27. What a shrinking head start means for the moats, margins, and IPOs of OpenAI and Anthropic.
OpenAI introduces GPT-Live, a full-duplex voice AI that listens and speaks at once and delegates deep reasoning to a separate model. A look at the three-way race with Google and xAI over the voice interface to AI.
OpenAI releases GPT-5.6 as Anthropic resets Claude's usage limits. With flagship performance converging, the AI race is shifting to usage-limit design and the competition to secure compute.
Credit card payment processor Zentoshin collapsed with 125.9 billion yen in liabilities and two decades of alleged fraud. What its failure reveals about early-payment finance and the hidden blind spot in a cashless economy.
At the July 2026 summit, Japan and India moved beyond investment into practical cooperation on economic security, defense, AI and energy. Not an alliance, but a working partnership built where interests overlap.
