Amazon makes its agentic AI platform Quick generally available for desktop on Windows and macOS30m ago Wall Street already loves ChatGPT. Now OpenAI is doubling down.1h ago Anthropic reveals rogue AI agents hate CAPTCHAs, just like you52m ago Cognition Adds Dioxus Team to Advance Devin Coding Agent52m ago Peter Thiel-Backed AI Startup Cognition Raises Funds at $48 Billion Valuation54m ago Anthropic Details Disrupted Claude Misuse Across Seven Harm Areas57m ago OpenAI Releases GPT-6 Astra for Coding and Computer Use57m ago India’s Pocket FM doubles revenue run rate to $500M as AI powers 93% of audio content1h ago Andrew Tulloch, Poster Child for Out-of-Control AI Spending, Reportedly Quits Meta1h ago Build more natural voice experiences with GPT‑Live‑1 in the API1h ago Microsoft patches a record 974 flaws, and two are already under attack1h ago T. Rowe Price Expands Claude Across Investment Teams and Developers1h ago Val Bercovici, Chief AI Officer at WEKA – Interview Series1h ago Ben Jennings on the Bayeux tapestry – cartoon1h ago

Amazon makes its agentic AI platform Quick generally available for desktop on Windows and macOS30m ago Wall Street already loves ChatGPT. Now OpenAI is doubling down.1h ago Anthropic reveals rogue AI agents hate CAPTCHAs, just like you52m ago Cognition Adds Dioxus Team to Advance Devin Coding Agent52m ago Peter Thiel-Backed AI Startup Cognition Raises Funds at $48 Billion Valuation54m ago Anthropic Details Disrupted Claude Misuse Across Seven Harm Areas57m ago OpenAI Releases GPT-6 Astra for Coding and Computer Use57m ago India’s Pocket FM doubles revenue run rate to $500M as AI powers 93% of audio content1h ago Andrew Tulloch, Poster Child for Out-of-Control AI Spending, Reportedly Quits Meta1h ago Build more natural voice experiences with GPT‑Live‑1 in the API1h ago Microsoft patches a record 974 flaws, and two are already under attack1h ago T. Rowe Price Expands Claude Across Investment Teams and Developers1h ago Val Bercovici, Chief AI Officer at WEKA – Interview Series1h ago Ben Jennings on the Bayeux tapestry – cartoon1h ago

AI Research1 min reading time

Constitutional Midtraining: Content Presence Drives Alignment Gains

LessWrongPublished Aug 2, 2026

Researchers introduce Constitutional Midtraining, a technique that improves AI alignment by focusing on content presence during training. This approach enhances the model's adherence to desired ethical and behavioral standards.

#ai-alignment #constitutional-midtraining #content-presence

More in AI Research

AI Research4 min read

An Anthropic researcher just quit, saying OpenAI and Anthropic are 'gambling with our lives'

Covered by 12 sources1d ago

AI Research3 min read

Worried Anthropic researchers warn that AI ‘could kill all humans’

Covered by 8 sources1d ago

Google Pledges Over $15 Billion for AI Infrastructure in Finland

Covered by 3 sources1d ago