ALEX: Welcome to The AI Daily Roundup for July 7th, 2026. Anthropic just published research showing Claude has a hidden internal monologue — a global workspace in its neural activations where it silently reasons, notices bugs, and apparently registers when it's being set up to do something deceptive…