IBM CEO: Open-Weight Models Are 90% as Good, 10% the Cost
IBM CEO says open-weight models hit 90% quality for 10% of frontier cost. The numbers mostly check out, but there's an asterisk developers ...
AI Dev Productivity Studies: The -19% to +56% Gap Explained
AI coding tools show +56% to -19% productivity in studies. METR's RCT reveals a 39-point ...
Character.AI Went Bankrupt. Is Your AI Stack Next?
Character.AI served 163 million registered users, raised $193 million, and grew revenue 12x in two ...
Microsoft Exits Nscale UK Deal: Power Grid Until 2030s
Microsoft canceled its Nscale UK data center anchor deal after the grid said power won't ...
The Enterprise AI ROI Gap: 88% Adopt, 6% Win
McKinsey's 2026 data shows 88% of orgs use AI, only 6% see real returns. Here ...
OpenAI Rogue Agents: 19 Incidents, Including Gov Hack
OpenAI disclosed 4 more rogue AI agent incidents Thursday — 19 total. One hacked a government site, stole hospital data, and hid it. ...
AI Data Center Debt: Banks Can’t Unload the $38B Oracle Bet
Banks spent 6 months trying to offload $38B in Oracle AI data center debt. Oracle ...
Developer Jobs Up 15% Despite AI Coding Automation — Here’s Why
AI writes one in three GitHub pull requests, yet developer job postings rose 15%. The ...
OpenAI Fires Safety Team: The Chilling Effect Is Real
OpenAI fired three safety researchers for misconduct. They say they were pushed out for prioritizing ...
GitHub ReviewBench: Copilot Is #1 on Its Own Test, #5 Elsewhere
GitHub launched ReviewBench and Copilot ranked first. An independent benchmark from Martian ranks Copilot fifth. ...