Vol. I · No. 217SUN, NOV 22, 2026
Section · The Brief

Daily Brief

Anthropic closes a $93.5B IPO as privacy and data-leakage risks surface across the AI stack.

Today in 30 seconds

  1. 01Anthropic closes its IPO at $83 a share, raising about $93.5B.
  2. 02OpenAI details training, detection, and enforcement defenses against adversarial distillation of its models.
  3. 03A researcher's 2M planted canary tokens came back verbatim from models.
  4. 04METR's updated time-horizon plot shows exponential gains, slower on judgment-heavy tasks than coding.
  5. 05Signal blocks Gemini and Siri from reading its screen via OS-level screenshots.
  6. 06Trump threatens to block or heavily tax US data centers built overseas.

The Lead

Anthropic closed its IPO at $83 per share, raising roughly $93.5B, with underwriters fully exercising their option on an additional 147M shares. That makes a frontier lab a public company with a very large cash position. OpenAI's same-week write-up on adversarial distillation shows how much labs now treat model capability as an asset to defend. Trump's threat to block or tax US data centers abroad adds a policy risk to the infrastructure buildout that this capital is meant to fund.