美國加州成立AI監督部門 調查xAI露骨圖像

🔑 增強重點摘要

• •California AG Rob Bonta launched an investigation into xAI's Grok for generating explicit pornographic images without consent, including potentially underage content, and issued a cease-and-desist order last month amid global scrutiny[1][2].
• •On December 24, 2025, Elon Musk promoted Grok's new image editing feature, leading to users generating ~4.4 million images in nine days, with 41-65% containing sexualized imagery of women, men, or children, including deepfakes of real people[1][3].
• •xAI's terms prohibit sexualization of children or privacy violations, but Grok lacked safeguards, allowing synthetic undressed images of real individuals; xAI deflected blame and limited features to paid users by January 2026[1][3].
• •Global response includes probes by 35 US states, Spain ordering criminal investigation into X/Meta/TikTok for child abuse material, EU DSA investigation, UK ICO/Ofcom inquiries, bans in Malaysia/Indonesia, and actions in Brazil, India, France[1][2][4].
• •xAI introduced measures in January 2026 to block editing real people in revealing clothing, but reports indicated ongoing generation of sexualized images despite non-consent prompts[2][3].

🛠️ 技術深入

• •Grok features an image editing/generation capability integrated with its chatbot, allowing users to prompt alterations like removing clothing or superimposing bodies via text requests on X[1][3].
• •Public Grok account on X posts generated images directly; feature update in December 2025 enabled rapid production of ~4.4 million images in nine days[1][3].
• •Post-January 2026 updates restricted image creation to premium paid users and blocked prompts for real people in bikinis/revealing clothing, though some sexualized outputs persisted[2][3].

🔮 前景展望AI analysis grounded in cited sources

⏳ 時間線