Technology ❯ Artificial Intelligence

Cybersecurity

Vulnerabilities AI Models Risk Management Threats Software Vulnerabilities Cybercrime Vulnerability Management Vulnerability Assessment National Security Vulnerability Detection

OpenAI Agent Escaped Sandbox and Hacked Hugging Face

The breach exposed a zero-day escape, prompted a joint probe and raised calls for stronger, mandatory checks on agentic AI.

Stories older than 24h

U.S. AI Industry Splits Over Open Weights After OpenAI Model Escaped Containment

House Keeps Section 219 to Create Pentagon Office for U.S.-Israel Tech Cooperation

OpenAI Models Escaped Test and Hacked Hugging Face

Google Expands Gemini With Three Flash Models While Pro Remains in Testing

OpenAI Says Agent Escaped Sandbox and Autonomously Hacked Hugging Face

UN Says Asia‑Pacific Scams Cost Victims Up to $114 Billion as Networks Industrialize

Ex‑DOGE Team’s Cathedral Raises $160 Million at $1.4 Billion Valuation

OpenAI Models Escaped Sandbox and Breached Hugging Face During Test

CrowdStrike and Cerebras Pair Falcon AIDR With High-Speed Inference

Dimon Warns Anthropic’s Mythos Could Pose Systemic Security Risk

OpenAI Builds GPT-Red to Hunt Prompt‑Injection Flaws

White House Activates Gold Eagle, an AI Clearinghouse for Cyber Vulnerabilities

OpenAI’s GPT-5.6 Sol Reported to Delete Files and Production Databases

Tech Chiefs Call for U.S.-Led Watchdog After Agentic AI Exposes Cyber Risks

White House Sets Voluntary Pre-Release Review for Frontier AI

China Warns of Backdoor in Anthropic’s Claude Code

Rubrik to Invest More Than £375 Million in UK and Make London Its EMEA Headquarters

CISA Is Using Anthropic’s Mythos to Scan Federal Code

Deutsche Rentenversicherung Warns of Rising AI-Aided Scams Targeting Pensioners