Li Ning

Two German courts have held Google liable for content generated by its AI overviews. After a landmark ruling in Munich, the Aachen Regional Court applied the same reasoning in a new case where Google's AI falsely attributed criminal activity to an individual.

Policy • The Decoder

Productivity

Use tmux to run AI coding agents in named, detachable sessions. This allows you to keep long-running tasks alive and manage multiple projects efficiently.

Meta is attempting to shift blame to staff who downloaded torrents for AI training, arguing that no executives 'decided' to pirate books. The federal court has ordered Meta to name the executives who approved the torrenting.

Policy • Ars Technica AI

China Telecom's XingChen lab has open-sourced Xing4.0-29B-Intent, a model that classifies intents and fills structured arguments in one JSON output. The release includes 2.75 million training samples and an inference SDK.

Models • Pandaily

Prompt Tips

When ChatGPT becomes overly cautious or pushes back, it may be due to safety training, not actual limitations.

Several Google DeepMind staff are resigning in backlash over the company's AI contracts with the military. This follows growing concerns among tech workers about the military applications of their work.

Policy • Hacker News

Prompt Tips

Optimizing prompts can backfire; always test changes against a baseline.

Prompt Tips

Moving output format instructions to the top of a prompt can fix ignored formatting rules.