#llm

2 signals tagged with this topic.

Why AI Agents Approve Their Own Bad Work (And How to Fix It)

Anthropic just admitted that Claude approves its own mediocre output. Their fix — borrowed from GANs — separates the agent doing work from the agent judging it. Here's how adversarial evaluation changes everything for agent systems.

March 26, 20268m read

Meta's Ghost Patent Isn't About Death. It's About Who Owns Your Digital Persona.

Meta patenting an AI that keeps posting after you die sounds dystopian. It is. But the dystopia isn't the ghost posting — it's what the patent reveals about who actually owns your behavioral data and what they intend to do with it.

February 23, 20267m read