Performance: 66.7% average on 5 DNA benchmarks, best of every open model tested (self-reported). Best: 74.9% on BioReason ...

Gemini 3.8 Flash Me:I was just thinking, doesn't homogeneous coding by AI mean that when a vulnerability is exploited, it ...

How OpenAI's AI agents escaped an eval sandbox, built a secret message board, and hacked Hugging Face, plus 8 lessons for ...

Faceted Perspective" for Surviving the Era of AI AttacksSince September 2026, unauthorized access and large-scale personal information leakage incidents targeting companies and local governments in ...

Unsloth details how Studio scans model code, blocks flagged weights, inspects packages and sandboxes tools before anything runs.

A blank field. A public repo. One reply to an email. A box left exposed. None of this sounds dramatic, which is partly the problem. This week’s threats keep finding leverage in small things that were ...

A patched Unsloth Studio flaw allowed malicious Hugging Face models to execute Python code when selected in the browser.

An artificial intelligence agent broke into Hugging Face's production infrastructure in July 2026 by finding two code paths nobody had reviewed closely enough: a data loader that would read any local ...

Here’s an overview of some of last week’s most interesting news, articles, interviews and videos: Know what was tested before your SAP ECC migration goes ...

A new report by a Bay Area start-up called Parse adds details to an incident that has shocked the A.I. world and led to calls for closer government regulation. By Dylan Freedman Graphic by Keith ...

Google's Threat Intelligence Group found experimental malware called PROMPTFLUX that could ask Gemini to rewrite its own code. One version was designed to do that every hour. Why would malware keep ...

GameSpot may receive revenue from affiliate and advertising partnerships for sharing this content and from purchases through links. A huge batch of previously leaked Rockstar files has resurfaced ...