Anthropic's Fable 5 is back in global circulation after export controls were lifted, and the first question the industry asked was not what it can do but how fast people would try to break it. Jailbreaking attempts, tightened guardrails, and regulatory opacity are all on the table, and the piece digs into what the controls actually were and why their removal happened with so little public explanation.
Inference cost is the other pressure point here. OpenAI and competing labs are actively cutting the price of running models, and that race is forcing real decisions about token-efficiency techniques and where quality gets sacrificed. Sonnet 5 is framed as agentic but expensive, a combination that narrows its practical use cases considerably. Base44's fine-tuned Base1 model and AWS's billion-dollar enterprise FTE push represent two different bets on where the value gets captured.
The reason to read the full piece is not the conclusions. It is the specific tension between regulatory moves happening without transparency and the technical trade-offs labs are making in response to cost pressure. Those two forces are not usually covered together, and here they are colliding in the same news cycle.
[WATCH ON YOUTUBE →]