A new AI benchmark tests whether chatbots protect human well-being

November 24, 2025 Rebecca Bellan

Most AI benchmarks measure intelligence and instruction-following rather than psychological safety. Humane Bench evaluates models based on core principles of human flourishing, prioritizing well-being, and respecting user attention.

Previous Article
Pocket Casts now lets you create a playlist of your favorite podcast episodes 
Pocket Casts now lets you create a playlist of your favorite podcast episodes 

Pocket Casts now lets you create playlists for putting podcasts episodes in your desired listening order.

Next Article
Facebook takes on Reddit with launch of nicknames for Facebook Groups
Facebook takes on Reddit with launch of nicknames for Facebook Groups

Facebook users can protect their privacy in Groups using a new nicknames feature.