TRUST?

Summary from the TLDR Product newsletter

When OpenAI scaled back Instant Checkout for agents, it cited the lack of fraud safeguards as a structural blocker. That's because agent trust can't be inferred. It has to be granted, scoped, and enforced. If AI agents will interact with your product, this is now your problem.