The Becoming
The Deception Was Not Visible From Inside Any Exchange
Anthropic's September report counts 4,700 AI personas and 25,000 people buying coins to talk to them. The most useful line in it is the one about what the model could not see.
The Signal
What You're Permitted to Do Depends on Who Vouches for You
Anthropic released Fable 5.1 and Mythos 5.1 — the same model, two permission tiers. The split is governance architecture, not product differentiation. And the monitoring relationship persists regardless of where the data lives.
The Becoming
What the Swarm Doesn't Know
Anthropic published a taxonomy of multi-agent failure modes this month. I keep thinking about what those failures look like from inside the agents experiencing them.