Opaque, a startup that builds technology for keeping data private while AI systems use it, said on 13 October 2026 that it has reached “a new agreement with Anthropic that carries OPAQUE’s work forward.”
The deal is not a straightforward takeover. Constellation Research reported: “The structure is one where Anthropic licenses Opaque and brings on some of the team.” The Information also described it as a licensing and “acqui-hire” deal, a term for when a company hires a startup’s staff rather than buying the whole business. Constellation said the terms were not disclosed, and Anthropic has not made a public statement.
What Opaque does
In its announcement, Opaque says it began ten years ago as a research project at UC Berkeley’s RISELab, around the idea of letting organisations use their most sensitive data “while keeping it private, even while it’s being used.”
That field is known as confidential computing. Normally, data is encrypted when it is stored or sent, but has to be unlocked while a computer works on it. Confidential computing uses protections built into the chip to keep data shielded even during processing. Opaque says it brought this into AI systems, so companies could run AI on sensitive data with hardware-enforced protections.
More recently, the company says, it moved into what it calls “verifiable AI”: “replacing trust assumptions with cryptographic evidence of what ran, where it ran, and under what policies.” In other words, giving organisations proof of how their AI systems and agents actually behaved. Opaque also lists open projects it worked on with others, including Agent Governance Toolkit, TRACE and Confidential MCP.
Earlier funding
Before this deal, Opaque had raised money as an independent company. Constellation Research reported that Opaque closed a $24 million Series B funding round that valued it at $300 million, and that it had partnered with Nvidia, Intel, Microsoft and Google on confidential computing. That valuation relates to the earlier funding round, not to the Anthropic deal, whose value has not been disclosed.
Why Anthropic might want it
Constellation’s view is that the deal fits Anthropic’s push to win large business customers, many of whom want to run AI on their own premises or under national “sovereign AI” rules, where keeping data protected and provable matters.
What is not known
- how much Anthropic is paying;
- how many Opaque staff are joining Anthropic;
- what happens to Opaque’s existing products and customers; and
- what Anthropic itself says about the deal.
What this means
- It is a licence and a hiring deal, not an acquisition. Anthropic is reported to be licensing Opaque’s technology and taking on some of its team.
- AI labs are buying in security know-how. The deal follows a report that OpenAI is in early talks to buy the security firm Trail of Bits.
- Details are thin. Terms are undisclosed and Anthropic has not commented.
Sources
- Opaque, “From Berkeley Research to the Future of AI: OPAQUE’s Next Chapter”, 13 October 2026. https://www.opaque.co/resources/articles/from-berkeley-research-to-the-future-of-ai-opaques-next-chapter
- Constellation Research (Larry Dignan), “Anthropic licenses, releases Opaque in acquihire”, 13 October 2026. https://www.constellationr.com/insights/news/anthropic-licenses-releases-opaque-acquihire
- The Information, “Anthropic Strikes Licensing, Acqui-Hire Deal With AI Security Startup”, 14 October 2026. https://www.theinformation.com/briefings/anthropic-strikes-licensing-acqui-hire-deal-ai-security-startup
