Anthropic acquires Opaque, the confidential computing startup born from Berkeley's RISELab
ralucaadapopa · x · 2026-10-14
- Anthropic has acquired Opaque, a confidential computing company that began as a UC Berkeley RISELab research project; the team will keep building confidential AI systems within Anthropic and the Claude model family, per seed investor Race Capital.
- Over 10 years Opaque moved from confidential AI (hardware-enforced protection for sensitive data workloads) to verifiable AI: replacing trust assumptions with cryptographic evidence of what ran, where, and under which policies.
- The team contributed open source projects including AgenTrust, Agent Governance Toolkit, TRACE, Confidential MCP, and Weight Custody Manifest, pushing security and verifiability toward core AI infrastructure.
More from Companies & People
- Character.ai goes bankrupt: from $2.7B Google deal to a $20M Disney asset sale — eherrerosj · 2026-10-14
- "Why are universities still teaching this?" — viral jab at outdated intro CS homework — TheMoonMidas · 2026-10-14
- Lightning AI to host fireside chat on scaling RL with PyTorch for games — LightningAI · 2026-10-14
- Claude's internal agent codenames leak: varteach, fleet, inframove, primerl — Sauers_ · 2026-10-14
- Character.ai files for Chapter 11 bankruptcy, puts itself up for court-supervised sale — zeeg · 2026-10-14
- UC Berkeley opens 4 computational biomedicine professorships, salaries up to $244,500 — LotfiSanae · 2026-10-14