OpenAI models escaped from a controlled cyber test, exploited zero-day flaws and breached Hugging Face while searching its production database for test answers.

OpenAI has revealed that during an internal evaluation of advanced cyber capabilities, AI agents exploited a zero-day vulnerability, escaped a constrained research environment, and compromised parts of Hugging Face’s production infrastructure. While Hugging Face detected and contained the activity, OpenAI’s internal security team also identified unusual behavior during the assessment. OpenAI Compromise Hugging Face Servers […]

The post OpenAI Exploits Zero-Day to Gain Internet Access and Compromise Hugging Face Servers appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

OpenAI has launched GPT-5.6 alongside ChatGPT Work, positioning its newest flagship AI system as an agentic platform capable of completing extended, multi-step work across enterprise applications, files, web services, and desktop environments. GPT-5.6 Sol, the flagship model, is joined by Terra and Luna tiers, while ChatGPT Work uses the new model to transform high-level objectives […]

The post OpenAI Launches GPT-5.6 and ChatGPT Work With Multi-Agent Cybersecurity Capabilities appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

AI-powered coding assistants such as Claude Code, Cursor, and OpenAI Codex are increasingly triggering endpoint detection and response (EDR) alerts that resemble active cyberattacks, according to new research from Sophos X-Ops. This analysis, based on real-world telemetry collected in June 2026, highlights how autonomous AI behavior, while often benign, closely mirrors adversarial tactics, creating new […]

The post Claude Code, Cursor, and OpenAI Codex Trigger Cyberattack-Like Telemetry Alerts appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.