Security leaders are being pitched a wave of new “AI security” tools. A nine-figure raise offers one signal about which risk the market considers real enough to fund.

Alice, the vendor formerly known as ActiveFence, announced a $140 million round on Aug. 25, 2026, led by Apax Digital Funds, which takes a board seat. The deal brings total funding to $280 million, with participation from Samsung, cybersecurity vendor SentinelOne, insurer Phoenix, and existing backers including CRV and Norwest.

The mix matters more than the number. When a security vendor and an insurer both back the same AI-defense platform, it points to institutional conviction that model-level guardrails, prompt-injection defense, and agent behavior monitoring are becoming a distinct category rather than features bolted onto existing stacks.

That thesis is investor framing, not settled fact. Apax partner Patrick Kane argues the “AI platform shift is opening a rapidly growing attack surface that will only widen as enterprises roll out agents.” Treat that as an interested party’s market bet, not verified research.

For buyers, the due-diligence gap is the story. Alice reports ARR “approaching $100 million,” AI-specific growth above 500%, protection for “more than 3 billion people,” and work with “8 of the 10” leading model labs. All are company-supplied and unaudited. A reported ~$800 million valuation traces only to anonymous sources.

The practical move is to treat funding as a signal about the category, not about the vendor. If AI security is becoming its own budget line, decide now which controls belong there (runtime guardrails, red-teaming, abuse detection) and which stay in existing identity, data, and application security programs. Then hold every vendor in the space to the same test: named references, independent evaluation results, and metrics you can audit yourself.