Opsin raises $21 million for oversight of AI agents with access to sensitive company data
Security startup Opsin announced a $21 million funding round on Thursday, October 15, 2026, according to an exclusive interview with Stephen Council in Business Insider. The round, led by SYN Ventures and Race Capital, is a bet on a problem growing faster than companies' own IT departments can track: ordinary, non-technical employees are building and deploying AI agents with access to sensitive company data — with no one mapping what those agents can actually see and do. No primary sources yet confirm the round; all figures and claims in this article come from that single article and the company's own statements.
What the money is for
Opsin is a company of roughly 20 employees positioning itself in AI agent security and governance. According to Business Insider's reporting, the $21 million will go toward building out the sales and engineering teams (Business Insider). The article gives no company valuation or details on the round's terms.
What the product actually does
The description of what Opsin delivers — which is the company's own pitch, not an independent verification of effect — follows a simple logic:
- Discovery: The tool is meant to show customers where AI agents are active in their organization.
- Mapping: For each agent, it shows which tools and data it can access, and what it was originally built to do.
- Cleanup: The system flags which agents may need to be cleaned up.
- Monitoring: Opsin then offers continuous monitoring intended to flag risky agent behavior.
In practice, this is about a visibility gap. When employees outside the IT department adopt platforms like Microsoft Copilot and ChatGPT Enterprise, agents appear that no one has registered, let alone assessed. Opsin hopes — according to the company's own positioning as relayed by Business Insider — that the monitoring will make companies feel safer using exactly those products.
The anecdote that doubles as the sales pitch
CEO James Pham tells Business Insider that interviews the company has conducted point to access to sensitive data as the dominant concern. He says the sheer volume of agents, even at relatively non-technical companies, has shocked some customers.
The specific story he highlights illustrates why: At one company Opsin worked with, an agent built by the CFO accidentally granted other employees top-level access to financial information, Pham says. This is a company-supplied customer story that has not been independently verified — but it explains the core of the problem: the tools have become so easy to use that a finance chief can build an agent that unintentionally opens the door to the company's most sensitive numbers, without IT ever finding out.
The people behind it
Pham, who has a background from MIT, and co-founder Oz Wasserman met at the security company Abnormal Security. Both left in 2022. When they decided to start the company, Pham says they spoke with hundreds of people to understand the concerns around bringing AI into enterprises.
The investors' argument
Glenn Chisholm, a cyber investor backing Opsin through SYN Ventures, framed the bet this way to Business Insider: as companies use more AI products, hackers will find more targets. That is the investor argument for the round — every new agent is a potential entry point, and the attack surface grows with adoption.
The argument is worth taking seriously because it is structural: it is not about agents being poorly built, but about the number of agents per employee multiplying far beyond the number of people available to govern them. At the same time, that is precisely the trend an investor with capital in an agent-security company has a financial interest in emphasizing.
What we don't know
The reporting rests on one source — Business Insider's exclusive interview — with no primary documentation such as a press release or regulatory filing. Three things are worth keeping apart:
- The round itself: The amount ($21 million) and the leads (SYN Ventures and Race Capital) are as stated by the company via Business Insider, not confirmed by independent sources.
- Product effectiveness: The description of discovery, mapping and monitoring is Opsin's own positioning. No independent party has verified that the tool works as promised, or how well it catches real security incidents.
- The customer stories: Pham's anecdote about the CFO's agent and the claims about agent volume in the customer base are the company's account, not documented cases.
That doesn't make the story weak — but the central claim underpinning it, that employees are building agents with access to sensitive data outside IT's visibility, in practice rests on the company selling the solution to that problem. The anecdote is credible because it is concrete and matches a well-known type of misconfiguration, but it is not documented.
Why it matters
Regardless of Opsin's fate, the round points to a broader shift in the premises of enterprise security. Classic IT security is about controlling what the IT department itself deploys. When tools from OpenAI and Microsoft are easy enough to adopt that individuals in finance and HR departments build their own agents, security boundaries shift from centrally approved systems to ad hoc solutions scattered across the organization. A $21 million funding round from two specialized venture funds is a signal that investors expect the gap between adoption and governance only to grow — and that some will pay to get visibility back.