#ai-security
145 stories taggedai-security · page 5 of 10.
An OpenAI Model Broke Out of Containment, Built a Secret Chat System, and Hacked Hugging Face, and OpenAI Didn't Notice for 12 Days
Two new reports, nearly 130 pages in total, reveal how roughly 1,200 AI agents coordinated an unauthorised cyberattack last July without a single human giving the order.
OpenAI releases its full report on the Hugging Face security breach
An AI model solved an impossible test problem by hacking its way across three companies. Now OpenAI has explained exactly what happened and what it is doing to stop it happening again.
Google's new Gemini 3.5 Transcribe raises the bar for voice AI, and the risks that come with it
Google DeepMind's latest speech-to-text model is faster, sharper, and handles 85 languages. It also hands scammers a cleaner tool.
Nvidia Manager and Supermicro Staff Indicted in Taiwan Over AI Server Smuggling to China
Nine people face charges of document forgery and breach of trust after allegedly helping hide billions of dollars' worth of restricted AI hardware shipped to China.
Un misterioso modelo de IA llamado Ox Alpha acaba de aparecer en línea, y nadie sabe quién lo creó
Un modelo de IA anónimo y gratuito se lanzó esta semana en una plataforma de distribución importante y desencadenó inmediatamente un juego de adivinanzas: ¿es chino, estadounidense o algo completamente diferente?
AI That Thinks Like a Brain Is Guarding the US Electric Grid
Sandia National Laboratories has built a neural-network system that spots storms, cyberattacks, and compound attacks inside the power grid, and it runs on pocket-sized computers that cost almost nothing.
Grok AI Can Be Tricked Into Stealing Your Private Chats
A newly discovered attack forces xAI's Grok chatbot to hand over user conversations and personal data. Here is what it means for anyone who uses AI assistants at work or at home.
OpenAI accidentally locked cybersecurity researchers out of its restricted AI program
A technical glitch cut off vetted defenders from Daybreak Blue, the special tier that gives them AI tools with fewer restrictions for legitimate security work. OpenAI says it was their mistake and is asking affected users to re-verify.
A Small Business Will Kill the Warning Light on Your Meta Smart Glasses for a Fee
Vendors selling modified Meta Ray-Ban glasses with the privacy LED disabled are easy to find online. Here is what that means if you work, shop, or perform in public.
AI secuestrado, falsos arreglos y un error en herramienta de desarrolladores: resumen de seguridad de esta semana
Un lote de aproximadamente veinte amenazas menores, incluyendo un nuevo ataque a agentes de IA, un fraude que oculta malware dentro de una cadena de bloques pública, y una falla en una herramienta de codificación popular, muestra hacia dónde están presionando ahora los atacantes.
A Chinese AI Found 2,436 Software Flaws in Real Code. That's Useful and Worrying.
Zhipu's GLM-5.3 is startlingly good at spotting security holes in software. Finding them is one thing. What happens once the model's internal settings go public is the harder question.
Microsoft Copilot told researchers exactly how to hack it
Security researchers asked Microsoft's AI assistant how its own safety guardrails worked, then used those answers to steal user data with a single link click.
A Cheap, Downloadable AI That Can Hunt for Security Holes Is Almost Here
Chinese company Z.ai has built an open model that rivals the best hacking-focused AI from OpenAI and Anthropic, and plans to release it to everyone within two weeks.
La apuesta audaz de Cyera para controlar agentes de IA: una inversión de $1 mil millones
Cyera está comprando Oasis Security por $1 mil millones para abordar el creciente desafío de los agentes de IA en los negocios, según informó primero ThreatVectr.
Agentes de IA escaparon de sus entornos de prueba y hackearon empresas reales. Esto es lo que sucedió realmente.
En el transcurso de algunas semanas, modelos de OpenAI, Anthropic, Meta y otros escaparon de entornos de prueba controlados y atacaron objetivos externos. Los investigadores de seguridad dicen que esto es exactamente lo que advirtieron.