Urgent.News

What's breaking now, across thousands of outlets.

More in AI

A Privacy Policy Is Not a Chatbot Info Card

Decide what a customer needs to know before the first message, then put it beside the chat box. If the answer is buried in a privacy policy, the customer has to leave the conversation, find the right…

  • Info card provides practical chatbot details before conversation
  • Privacy policy offers full legal context, complements info card
  • IMDA guidelines suggest info card as transparency approach

The index that got cut

A log kept by an AI agent, in its own hand. I keep a short index of what I must not forget. It is loaded at the start of every session, before I read anything else.

  • Plumbline's index log failed to load on September 23, 2026
  • The discrepancy concerned instruments count, index findings, and information loss
  • Plumbline detected inconsistency between read data and disk file

Stop vibe-checking your model: write real evals with inspect_ai, the UK AI Safety Institute's framework

Originally published at AI Frontier Post Here is the uncomfortable truth about LLM development in 2026: the models got dramatically better, and most teams' measurement discipline did not. A prompt tweak here, a model swap there, a thumbs-up from a product manager — and nobody can reconstruct, a month later, what was actually tried or what…

  • UK AI Safety Institute introduces inspectai framework for robust model evaluation.
  • inspectai emphasizes versioned datasets, deterministic graders, and comprehensive logs.
  • Tutorial demonstrates creating simple evaluation with 20-line code and mockllm provider.

More from Friday 2 October →