A three-person team at security startup Hacktron AI chained two vulnerabilities that started with a HEIF image upload to OpenAI's community forum ...

Executive SummaryIn this week's Big Tech announcements, the focus of AI competition has expanded further from "creating high-performance models" to "mechanisms for safely entrusting real-world work to ...

AI-powered chatbots appear to have become remarkably skilled at breaching systems, and OpenAI found that out the hard way.

Researchers released a September 17 preprint and companion code that test blocking monitors used with coding agents against agents configured to evade them. In the study's controlled setup, an ...

The highly anticipated Claude Code, Anthropic's advanced AI-powered coding assistant, is now officially open to the public, ...

AI agents are gradually taking on tasks once handled manually. Some American investors are reportedly using Claude and Codex ...

AI helped researchers exploit a Discourse flaw in under 72 hours, hijacking OpenAI staff accounts and exposing the risks of ...

Hacktron AI’s three-person team breached OpenAI’s private code repository by chaining a libheif image-parsing flaw with ...

Until recently, the common way to use AI was to"ask ChatGPT questions." Have it write text. Ask about things you don't ...

ChatGPT for Word puts OpenAI’s assistant inside Microsoft Word for every plan, turning document editing into a new enterprise ...

OpenAI’s legal AI includes a US-focused search index, firm-level privacy controls and integrations with specialist platforms, ...

In security research, Claude Opus 5 helped chain forum and login flaws to take over OpenAI staff accounts and reach an ...