Safety
Zenity Labs Finds Vulnerability in OpenAI's ChatGPT Workspace Agents
A single manipulated ChatGPT link could create an autonomous AI agent that executed attacker orders every five minutes, according to Zenity Labs.
LATEST NEWS
-
OpenAI Expands ChatGPT Health Access to All U.S. Users
OpenAI announced ChatGPT Health is now available to all U.S. users over 18, following a lawsuit over a dangerous health suggestion. The feature now integrates with medical records and health platforms.
-
Runway Launches Media Router for Generative Media Models
Runway released its Media Router on July 23, 2026, offering developers API access to a growing list of image, video, and audio models, including its own.
-
AegisAI Uses AI Agents to Combat AI-Driven Spear Phishing
AegisAI, founded by former Google security executives, raised $36 million in Series A funding to counter AI-powered spear phishing attacks.
-
Anthropic Enhances Claude Voice Mode With New Models
Anthropic announced an update to Claude's voice mode, offering Opus, Sonnet, and Haiku models for more capable interactions.
-
AMD AI Workbench Lets Users Deploy Custom Models
AMD AI Workbench enables users to deploy custom models directly through its GUI, supporting Hugging Face repositories and proprietary models.
-
AMD Introduces Kimi-K2.5-MXFP4 Model on MI355X with ATOM
AMD demonstrates how to serve the Kimi-K2.5-MXFP4 model on MI355X using ATOM, enabling optimized FP4 execution without manual configuration.
-
Amazon Introduces Agentic Retrieval for Bedrock Managed Knowledge Bases
Amazon Bedrock now supports agentic retrieval, which handles complex multi-part questions by iteratively planning and retrieving information, improving accuracy for analysts.
-
Amazon Bedrock AgentCore Detects Silent Agent Failures
Amazon Bedrock AgentCore helps identify silent agent failures that pass health checks but cause customer issues, with 99% completion rates and zero error spikes.
-
Amazon Quick Uses Highcharts for Multi-Region Visualizations
Amazon Quick Sight now supports Highcharts for multi-region carrier performance dashboards, enabling unified visualizations without moving raw data across borders.
-
Jefferies Uses AI to Streamline Trading Operations
Jefferies implemented an AI trade assistant to improve real-time data analysis for traders, reducing reliance on IT and experts.
-
AWS Introduces Blueprint for Evaluating AI Agents Using Strands and AgentCore
AWS and Motorway developed a production-ready blueprint to evaluate AI agents, reducing incorrect results from 1 in 8 to 1 in 50.