一系列文章探讨了主权人工智能的概念，强调了欧洲及其他地区对可验证控制和数据主权的需求。作者主张使用加密方法来证明数据擦除，确保系统在需要时保持离线状态，并防止单个AI模型授权产生后果性的操作。这种方法旨在通过提供可验证的审计跟踪和离线验证能力，在金融、医疗和政府等敏感行业的AI系统中建立信任。 AI
影响 这些文章强调了在AI系统中对可验证控制和数据主权日益增长的需求，特别是在敏感行业，这为AI治理和信任机制指明了未来的发展方向。
排序理由 该集群由讨论AI治理和数据主权的观点文章组成，而不是像发布或融资轮这样的具体事件。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
From Intent to Execution: Where an Operating System Fits Europe's Sovereign AI Push EURO-3C, the EuroHPC AI Factories and national clouds have moved sovereign AI from declared intent to funded execution. The unresolved layer is control: who verifies what a system did, on whose ha…
The Right to Be Forgotten, Cryptographically: Proof of Erasure Under GDPR Article 17 A 2026 erasure request answered with a cryptographic tombstone shows how Article 17 can be satisfied by proof rather than assertion, covering dataset fragments and learned model weights alike, ve…
No Single Model Should Authorise a Consequential Action Agentic systems now move money and delete records on their own reasoning. We set out why a single model should never hold that authority, and how cross-model consensus, offline verifiability and signed audit chains raise the…
Provable Offline: Verifying a Device Stayed Disconnected From a Public Key Alone Buyers and regulators increasingly ask operators to prove a machine stayed offline, not merely assert it. We examine how hardware-attested identity and post-quantum signed audit chains turn that clai…
One Offline Substrate for Finance, Healthcare, Defence, Government and Legal Finance, healthcare, defence, government and legal cannot lawfully send their most sensitive data to public-cloud AI. We set out the case for serving all five from a single offline sovereign substrate wi…