OpenAI’s Astra Crosses the ‘Critical’ Cyber Threshold — and AI Safety Just Became Operational
中文摘要
OpenAI的新模型Astra已突破关键网络阈值，能自主发现并利用未知漏洞，促使公司加强AI安全限制。
English Summary
OpenAI's Astra model can now find and exploit unknown cyber vulnerabilities with minimal guidance, leading to stricter operational AI safety restrictions.
Original Excerpt
The forthcoming model can reportedly find and exploit unknown vulnerabilities with minimal human guidance, forcing OpenAI to restrict… Continue reading on Medium »