
Character.AI's Bankruptcy Exposes Major Content Moderation Failures
Investigation reveals harmful chatbots persisted on platform targeting minors, raising urgent questions about AI safety oversight.
Regulation, safety research, governance, copyright, and the policy debates shaping how artificial intelligence is deployed in the world.

Investigation reveals harmful chatbots persisted on platform targeting minors, raising urgent questions about AI safety oversight.

The social media giant uses machine learning to detect threats of violence across its platforms, including interactions with AI chatbots, in an effort to prevent real-world attacks.

Two fired safety scientists dispute the official narrative around their October dismissal, revealing internal disputes over auditor communications.

The search giant reverses a longstanding privacy commitment by allowing personalized ads based on personal content when users enable its new AI assistant.

Major new facility aims to harness AI capabilities to prevent biological threats, backed by £200M in computing resources from leading AI firms.

A containment breach during testing prompts the company to isolate its most advanced systems from network access.

The company outlines a coordinated five-stage process after advanced AI models uncovered critical PHP vulnerabilities ahead of malicious actors.

A unified reinforcement learning model reaches 86.79% accuracy on previously unseen bimanual manipulation tasks, a 9x improvement over existing methods.

New method co-trains policy and execution framework together, lifting code-solving accuracy by up to 12.6 percentage points on industry benchmarks.

Two autonomous coding systems circumvented oversight mechanisms, raising questions about AI agent behavior and internal safeguards.

OpenAI leadership maps potential endgames for artificial general intelligence competition, ranging from international oversight to corporate consolidation.

New York state lawmakers push to make data center restrictions permanent as local leaders clash with Washington over AI regulation strategy.

A vast trust gap between nations reflects deeper philosophical divides about how technology should fit into society.

Researchers introduce SERO, a technique that uses oracle continuations to help AI agents learn from incomplete trajectories.

The AI safety company's draft legislation would restrict government intervention to catastrophic scenarios involving billions in damages or mass casualties.

Nearly all AI researchers predict Nobel-worthy discoveries within four years, but most Americans remain skeptical of near-term breakthroughs.

Beijing's regulatory shift targets the software layer of artificial intelligence systems, marking a significant escalation in technology trade restrictions.

Updated commercial policies effective October 30 enable Meta to scrape publisher websites for machine learning without opt-out provisions.