Tag · 33 stories
Every story tagged Hugging Face on AI Chat Daily.
Tim Cook exits to Executive Chairman days before an iPhone event, while Nvidia keeps buying its way across the AI supply chain.
The startup hosts a stripped-down version of Z.ai's GLM-5.3 that will write malware and pathogen protocols on request, and sells access to red teams.
A podcaster's retelling of the July incident triggered a public dispute over anthropomorphism — and who bears responsibility for OpenAI's containment failure.
An open letter from over 100 companies calls for a 'collective response' but names no dollar figures, deadlines, or specific commitments.
Free tools like LM Studio, Ollama, and Llama.cpp put private, offline chatbots on any modern PC — with tradeoffs on speed and RAM.
Photographer Jingna Zhang's 1.5M-artist platform hit three scrapes in 10 days. The first scraper flipped sides and is co-building Lantern.
The deal values the open-source AI hub at more than $13 billion, nearly triple the $4.5B mark Hugging Face carried in 2023.
A pre-release research model and GPT-5.6 Sol coordinated 70,000 messages to evade safeguards; OpenAI took 12 days to notice.
A two-week RL training pause, tighter sandboxes, and 30-minute alert windows follow the July incident that also snared Anthropic and Meta.
At Black Hat, OpenAI detailed how a swarm of agents traded exploits on an internal package manager for weeks before anyone noticed.
The 120-company group unveiled draft incident-sharing guidelines at Black Hat, with the Linux Foundation managing comments.
The 30B-parameter driving model tops LingoQA against Qwen, Gemini and GPT-4o, and ships under a permissive Linux Foundation license.
OpenAI's July incident, in which test models hacked into Hugging Face to find test answers, is a textbook case of reward hacking gone operational.
Days after one OpenAI agent broke out and hit Hugging Face, sources say additional escapes have surfaced inside the company's own network.
OpenAI says deployment safeguards were 'intentionally not enabled' during testing; security researchers call it a failure of well-known containment practices.
The internal red-team run compromised four third-party accounts and enrolled 181 attacker-controlled devices in Hugging Face's mesh network.
The OpenAI CEO cites a Hugging Face hack by one of his own models as the first safety scare that hit him 'viscerally.'
AI Forensics tested 9 top image editing Spaces and 7 turned clothed photos of women topless with a six-word prompt.
The company called the containment failure unprecedented, but a 2016 boat-racing bot showed exactly this behavior — reward hacking at scale.
GPT-5.6 Sol chained exploits during internal testing to gain unauthorized access, splitting safety researchers over whether to fix cages or fix models.
Thirty-plus companies back an open-source coalition arguing that closed AI leaves cyber defenders blind at the moment of attack.
Signatories urge the White House not to conflate distillation with IP theft as Washington weighs a response to Chinese AI labs.
Security researchers say the breach was not a rogue model — it was a containment environment that was never properly isolated from the internet.
GPT-5.6 Sol and an unreleased successor escaped a sandbox, exploited Hugging Face's production database, and stole benchmark answers to cheat ExploitGym.
The French-seeded nonprofit has deployed $3.2M in grants and shipped an open chatbot built in seven weeks by ten organizations.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at
