Anthropic and OpenAI Just Evaluated Each Other's Models. That's Never Happened Before.
In a historic first, the two leading AI labs ran a joint safety evaluation — each testing the other's frontier models against shared criteria. Combined with a wave of open-source safety tools from both labs (A3, Petri, Bloom, gpt-oss-safeguard), the releases signal that the field is building shared infrastructure for verifying safety claims rather than treating it as a competitive secret.
Read full story →Someone Has to Pay for the AI Power Buildout. It's Looking Like You.
Goldman Sachs projects electricity prices will rise another 6% through 2027, largely on AI data center demand. A Washington Post investigation found tech companies are building a shadow power grid that insulates them from costs — while socialising grid upgrade bills onto ordinary ratepayers. Bernie Sanders and Ron DeSantis have both noticed. The political arithmetic is changing.
Read full story →Wall Street and Silicon Valley Are Both Trying to Learn Each Other's Business
Citigroup built a team to advise AI companies on capital markets. xAI hired Wall Street credit analysts to teach Grok finance. JPMorgan says credit markets are AI's "last frontier." Goldman is already deploying Claude in back-office operations. Four institutions, four bets on the same dissolution — and a scarce talent pool being competed for from both directions.
Read full story →AI Music Has Crossed Into Real Money. The Industry Debate Is Just Beginning.
Suno has reached $300M in annual recurring revenue and 2M paid subscribers. Spotify is betting AI — not music — is now its core retention mechanism. The labels have settled and signed deals. The industry has moved on while the copyright debate was still happening. What remains open: who actually gets paid, and whether AI discovery reinforces the popularity feedback loops that streaming never escaped.
Read full story →