← BACK TO HOME SIGNALS

Signals

Short provocations on what is shifting in AI security and AI usage. Each signal makes a sharp claim, anchors it with the evidence, and opens the matching deep reference hub. Read one in two minutes; go deeper when something hits.

  1. Model Economics · May 23

    Enterprise AI cost is silently doubling. Three deployment patterns hide the markup.

    Inference is the line item every CFO watches. The 1.6x–3x compounders sit outside it — and nobody is auditing them.

    ↳ Model economics playbook Read signal →
  2. Enterprise Adoption · May 23

    Smart companies, dumb AI mistakes. What three high-profile rollbacks revealed.

    Air Canada, Klarna, Cursor. Different products, same failure shape — and it is not the model.

    ↳ Enterprise rollout patterns Read signal →
  3. Model Releases · May 23

    What changes in the next Claude model — and what stays broken.

    Bigger context, sharper agentic ergonomics, the same indirect-injection weakness. Capability gains do not retire threat surfaces.

    ↳ Claude Agent SDK threat model Read signal →