LIVE · refreshes every 20 min

updated Nov 6, 05:04 UTC

← 110010.art all stories

Hacker News (AI) ·27d ago

OpenAI cannot make AI safe on its own; arXiv paper analyzes safety challenges and author claims

Read the original at mikitabalesni.com→

Original headline: "OpenAI cannot make AI safe on its own [pdf]"

Coverage timeline

Oct 9, 05:16 UTC Hacker News (AI) lead source OpenAI cannot make AI safe on its own [pdf]

Related stories

OpenAI releases GPT-6 Cyber, a cybersecurity model for finding vulnerabilities and developing fixes within an authorized scope; OpenAI says it is better at not hacking other companies

OpenAI · 20h ago · source ↗

Anthropic launches Claude Duet, a new AI agent challenger to Muse and Dots, available as a standalone iOS/Android app and via Claude on the web.

The Verge AI · 15d ago · source ↗

Grok’s AI model overwhelmingly recommends Republicans while OpenAI’s GPT favors Democrats in election predictions.

TechCrunch AI · 12d ago · source ↗

OpenAI quietly renames data centers to “compute campuses”

The Verge AI · 2d ago · source ↗

OpenAI models come to Google Cloud via Gemini Model Garden and Google Cloud Marketplace

OpenAI · 15d ago · source ↗