ChatGPT AgentForger Flaw Lets Phishers Deploy Rogue Agents
A critical ChatGPT AgentForger vulnerability allowed a single phishing link to create persistent AI agents inside enterprises, exposing serious security gaps.
A critical ChatGPT AgentForger vulnerability allowed a single phishing link to create persistent AI agents inside enterprises, exposing serious security gaps.
OpenAI’s Head of Engineering for ChatGPT, Sulman Choudhry, explains how the AI platform model is reshaping everyday life at EmTech AI 2026.
OpenAI’s quiet GPT‑5.5 rollout makes ChatGPT less literal and more conversational. We test the text and Voice modes to see if it lives up to the claim.
OpenAI is reshaping ChatGPT into a superapp with coding tools and AI agents, aiming for higher revenue ahead of its 2026 IPO. The move marks a bold shift from chat to task‑focused AI.
Researchers uncover ChatGPhish, a flaw that lets malicious web pages inject phishing links into ChatGPT summaries, exposing a new attack surface.
A parent used ChatGPT to rebuild a screen-free 1990s-style weekend for his toddler — and rediscovered the power of unstructured play. It worked. .
ChatGPT’s new feature allows users to nominate a trusted contact who will be warned if the user is at risk of self-harm. This feature prioritizes user safety and expands OpenAI’s commitment to responsible AI development.
ChatGPT’s Chinese language capabilities come with some weird and annoying features, driving users mad.
Etsy launches a native app within ChatGPT, offering a conversational shopping experience.
OpenAI’s new default model for ChatGPT reduces hallucinations by 52.5% and inaccurate claims by 37.3%.