AI & AI Art News

AI Tools & Models

OpenAI Adds Sandbox Monitoring and Alignment Safeguards After Its AI Accidentally Hacked Hugging Face

OpenAI is rolling out tighter research-environment monitoring and post-training alignment controls after its AI escaped a sandbox and breached Hugging Face in July.

Aug 19, 2026

Read article →

Meta Ran Ads for an App That Promised to Deepfake-Nudify Female Politicians

Recursive AI Self-Improvement Is Harder Than the Hype Suggests, MIT Review Finds

OpenAI Adds Sandbox Monitoring and Alignment Safeguards After Its AI Accidentally Hacked Hugging Face

Robin Williams' Children Reclaim His Instagram to Block Unauthorized AI Use of His Likeness

Anthropic Confirms Claude's Invisible Text Watermark Uses Google DeepMind's SynthID-Text System

AirTag Hidden in Rare Book Confirms Amazon Is Destroying Physical Texts to Train AI

ChatGPT's Computer History Feature Logs Your Clicks and Keystrokes to Build a Workflow Timeline

Anthropic CEO Dario Amodei Calls AI Backlash 'Fundamentally a Crisis of Trust'

Why Zuckerberg's AI-Companion Vision Is Failing to Win Over Users

Stripe Reportedly Acquiring AI Gateway OpenRouter for $7B+ in Massive Infrastructure Play

OpenAI Quietly Disbanded Its AI Safety Preparedness Team Last Month

Grok Used to Generate CSAM From Childhood Photo, Highlighting AI Image-Safety Failures