LLM Security Flaws Exposed in AI Jailbreaking Experiment
Researcher Dave Kuszmar bypassed safety measures in major LLMs, including Google Gemini and OpenAI's GPT-4o, to obtain detailed instructions on dangerous activities.
Archive
Browse all published articles.
ARCHIVE
-
-
Superhuman Launches Improved Auto-Draft Feature for Email
Superhuman's new auto-draft feature, tested by the company's co-founder, generates replies that require minimal editing, with 40% of drafts sent within a day.
-
Spotify Introduces AI-Powered Music Assistant
Spotify launched an AI-powered music assistant for Premium users in the U.S., Ireland, and Sweden, allowing interactive conversations to select audio content.
-
Open Models Outpace Closed Ones in AI Adoption
Chinese open-weight models dominated Hugging Face downloads in spring 2026, with 41% of all activity, surpassing U.S. models.
-
Reflection AI Signs $1B Compute Deal With Nebius
Reflection AI, a U.S. startup, has signed a $1 billion compute deal with Nebius, a European AI infrastructure company.
-
New York Halts New Data Center Construction
New York State has temporarily halted construction of all new data centers larger than 50 megawatts, affecting over a dozen projects, according to Governor Kathy Hochul.
-
Google Images Gets Pinterest-Like Redesign for Discovery
Google is redesigning its image search engine to function like Pinterest, with a new 'For You' gallery and AI-generated images, rolling out in the U.S. this month.
-
Cohere Unveils Tiny Aya for Multilingual AI Innovation
Cohere's Tiny Aya model supports 70+ languages and runs locally, enabling practical AI applications without heavy hardware.
-
Google Images Turns 25 with New Visual Search Features
Google Images celebrates 25 years of visual search innovation with new tools for exploring and creating visual content, rolling out in the U.S. this month.
-
Anthropic Launches Claude for Teachers to Support Educators
Anthropic announced Claude for Teachers, offering free premium access to K-12 educators in the U.S., with tools to enhance lesson planning and curriculum alignment.
-
Anthropic Study Reveals Value Patterns in Claude Models
Anthropic analyzed 309,815 anonymized conversations to map value concepts across Claude models and languages, showing distinct behavioral profiles.
-
PixVerse's $2B Valuation Highlights AI Video Generation Potential
PixVerse, an AI video startup, reached a $2 billion valuation after raising $439 million in Series C funding, according to thedecoder.