Updates on Model Safety

Every AI story we track on Model Safety: 12 stories so far, each summarized in our own words and linked to the publisher that reported it.

Pulled from 125 sources

This month

Cybersecurity6 min read

AI’s ‘Abliteration’ Problem Is Bigger Than China

Gizmodo24d ago

The Morning Risk Report: Trump Defends Light-Touch AI Strategy After Industry Sounds Alarms

The Wall Street Journal25d ago

LLM & Text Generation1 min read

Astra 6.1 Pulled As Insufficiently Aligned

LessWrong26d ago

Scoop: Top AI companies probing tens of thousands of security incidents

Covered by 2 sources29d ago

Business & Enterprise5 min read

Open Weights, Life Sciences And The Risk Of Enterprise AI Dependency

ForbesSep 22

OpenAI Shares More Safety Incidents and Adopts New Rules for Reporting Them

Covered by 8 sourcesSep 16

Machine Learning3 min read

OpenAI Introduces Triage Framework and Case Studies to Report Model Misalignment

InfoQ (AI, ML & Data)Sep 18

LLM & Text Generation2 min read

How Value Induction Reshapes LLM Behaviour

Apple Research BlogSep 16

AI Research8 min read

Open-Source AI & Open Models Reading List

InterconnectsSep 11

AI Research7 min read

Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic

Hugging FaceSep 8

AI Research3 min read

Security Concerns Cause OpenAI to Halt Work on Astra Model

Covered by 2 sourcesAug 10

AI Research15 min read

Predicting model behavior before release by simulating deployment

Open AI NewsJun 16

That's everything we have on Model Safety right now