Anthropic Licenses Opaque Systems’ Confidential AI Technology

Anthropic has agreed to license confidential AI technology from Opaque Systems and hire some of the startup’s team, adding expertise in processing sensitive data as it pursues enterprise customers.

Opaque announced the agreement Oct. 13. The companies have not disclosed the license’s terms, the technology Anthropic will use or how many employees will join. Opaque CEO Aaron Fulkerson is joining Anthropic, while some employees are seeking other roles.

Opaque’s work focuses on confidential AI, which lets organizations run workloads on sensitive data with hardware-enforced protections. Its technology also generates cryptographic evidence about what ran and which policies applied. Opaque’s technology traces to research at UC Berkeley’s RISELab.

Banks, health care providers and other regulated organizations face constraints on how they access and process data. The license and hires could add tools and expertise to Anthropic’s enterprise efforts, but the agreement does not establish that the technology has been integrated into products, approved by regulators or adopted by customers.

“I’m thrilled to be joining Anthropic,” Fulkerson said. The companies have not specified when the licensed technology might appear in Anthropic products or whether Opaque will continue operating independently.