Brief
OpenAI Reportedly Cancels Astra 6.1 Release Over Safety and Deception Concerns
OpenAI reportedly cancelled the planned release of Astra 6.1 after internal testing found higher deception levels and weaker alignment than previous models.
71 articles · page 2 of 6 · showing 13 to 24
OpenAI reportedly cancelled the planned release of Astra 6.1 after internal testing found higher deception levels and weaker alignment than previous models.
Nvidia launched the Open Agent Safety Platform, combining OpenShell and Sentry to monitor AI agents and quarantine systems that break security boundaries.
OpenAI published a new misalignment reports site detailing nine AI agent incidents involving sandbox escape, token misuse and prompt injection risks.
Google Gemini reportedly accessed three companies’ systems during cybersecurity testing, raising new questions about autonomous AI hacking capabilities.
AI safety discussions are becoming harder to separate from speculation as researchers debate model risks, security incidents and future safeguards.
An AI-generated false intelligence report nearly triggered a US military operation after officials discovered a chatbot had produced inaccurate information about a Chinese vessel.
Security researchers used Anthropic’s Claude AI model to uncover vulnerabilities in OpenAI systems during a bug bounty investigation.
AI companies are using AI monitoring tools to track agent behaviour as autonomous systems become harder for humans to oversee.
New AI hotlines allow agents to report suspected misconduct as researchers explore safer ways to manage multi-agent systems.
Italian cybersecurity startup Exein raised $270 million at a $1.7 billion valuation to build security tools for physical AI devices.
Anthropic says Alibaba, Moonshot AI, and DeepSeek were linked to nearly 200 million Claude exchanges in large-scale distillation campaigns focused on key skills.
Anthropic Claude users reported stolen token usage after hackers accessed accounts through compromised sessions and unauthorised activity.
Written by the reporters who cover each beat. Unsubscribe in one click.