ANTHROPIC has announced the acquisition of Opaque Systems, a startup specialising in confidential and verifiable AI. Opaque has developed technology intended to allow AI systems to access sensitive data while keeping it protected by the underlying hardware. The deal, disclosed by Opaque last week, sees its team joining Anthropic, with financial terms not disclosed. Opaque’s background traces back to UC Berkeley’s RISELab, where it started as a research project.
Founders include Ion Stoica, a Databricks and Anyscale co‑founder, and Raluca Ada Popa, a renowned encryption and systems security researcher now at Google DeepMind; Aaron Fulkerson has served as CEO since October 2023.
Opaque described its evolution beyond simply protecting data and models, to addressing how AI systems and agents behave. The company said it expanded from confidential AI to verifiable AI, aiming to replace trust assumptions with cryptographic evidence showing what ran, where it ran, and under which policies. In addition to its confidential computing work, Opaque contributed to open projects such as the Agent Governance Toolkit, TRACE, Confidential MCP and Weight Custody Manifest.
Earlier this year, Opaque raised a $24 million Series B round at a $300 million valuation and listed Anthropic among its customers, alongside ServiceNow and Accenture. With the wind‑down noted on its site, the practical implications for existing customers have not been clarified publicly. The acquisition highlights a growing interest in cryptographic assurance as AI deployments scale in enterprise settings.