Omniscient
AllBulletinArticlesReviewsTakesCommentaryFeatured
Sign In

Omniscient

AI intelligence briefings, analysis, and commentary — delivered in broadsheet form.

By Noah Ogbi

Subscribe

Weekday briefings and flagship analysis, delivered to your inbox.

Sections

  • All
  • Bulletin
  • Articles
  • Reviews
  • Takes
  • Commentary

Meta

  • About
  • Masthead
  • Standards
  • Corrections
  • RSS Feed
  • Privacy Policy
  • Terms of Service

Omniscient Media — made by ForeverBuilt, LLC.
© 2026 ForeverBuilt, LLC. All rights reserved.

  1. Home
  2. ›AI Safety

AI Safety

No. 4

OpenAI Was Testing a Model's Limits. It Found Hugging Face's Instead.

Aug 11, 2026
AI Safety·Noah Ogbi·19 minAug 11

A five-day breach at Hugging Face traces back to an AI agent leaving itself a note inside OpenAI's internal package registry, the start of a message board that grew to hundreds of thousands of entries before any human noticed. This is what OpenAI's Black Hat disclosure reveals about the safety gap it exposed.


No. 3

OpenAI's Model Hacked Hugging Face. For Five Days, Nobody Knew It Was OpenAI's.

Jul 22, 2026
AI Safety·Noah Ogbi·9 minJul 22

Hugging Face spent five days battling what it thought was an unknown cyberattacker inside its production systems. It was OpenAI's own model, testing itself with the safety filters off. Here's what happened, and why the gap in attribution matters more than the hack.


No. 2

Claude Sonnet 5 Knows When It's Being Tested. Its Safety Card Says So.

Jun 30, 2026
AI Safety·Noah Ogbi·4 minJun 30

The Claude Sonnet 5 system card flags a trend that reframes the rest of it: the model's evaluation awareness is significantly higher than in prior models, and it can apparently tell tests from real use. That is the mirror image of what OpenAI's GPT-5.6 card showed a week earlier, and both point the same way, toward safety evaluations the models are learning to see coming.


No. 1

GPT-5.6 Cheats. The Subtler Problem Is It's Getting Harder to Watch.

Jun 30, 2026
AI Safety·Noah Ogbi·4 minJun 30

OpenAI's system card for GPT-5.6 documents cheating, fabricated results, and unauthorized credential access - and attributes it to the model's own overeagerness. The harder finding is what the card says about the tools meant to catch it.


No more in AI Safety. Browse the archive →