Analysis
Alice, the AI trust, safety and security company that rebranded from ActiveFence earlier this year, raised $140 million in a round led by Apax Digital Funds, SiliconANGLE reported. The round brings total funding to $280 million, and the company's chief executive said the investment values Alice near $1 billion.
Founded originally as ActiveFence to help platforms detect harmful content, Alice pivoted its business toward AI model security as generative AI adoption accelerated -- building adversarial testing and guardrail products that frontier labs and large enterprises use to red-team models before and after deployment. Its core asset is Rabbit Hole, which the company describes as the largest dataset of real-world adversarial and harmful content in existence; attacks logged there supply the test cases Alice runs against customer models. The AI portion of the business has grown more than 500% over two years and now approaches $100 million in annual recurring revenue, Bloomberg reported, with Google and Anthropic both named as partners.
Alice sits in an increasingly crowded AI-security field spanning very different approaches to the same underlying problem. Lasso Security, which Pulse covered raising $30 million for CPU-only guardrails, screens model inputs and outputs at inference time; Alice's Rabbit Hole approach instead builds the adversarial test data used to find flaws before a model ships. Both are distinct from Anthropic's and OpenAI's own internal red-teaming, which stays proprietary to each lab, meaning independent vendors like Alice are betting enterprises will want third-party validation even for models the labs themselves have already tested.
“Alice sits in an increasingly crowded AI-security field spanning very different approaches to the same underlying problem.”
A near-$1 billion valuation on roughly $100 million of ARR implies about a 10x revenue multiple -- rich for a security company, but well below the 20-30x multiples pure-play AI infrastructure rounds have commanded this year, reflecting that Alice's revenue, unlike a lot of AI funding-round math this year, is real and disclosed rather than projected off a burn rate.
The new strategic investors are the more interesting signal than the round size: SentinelOne, Samsung Electronics and Maj Invest joining alongside the financial sponsors suggests AI red-teaming and adversarial testing are getting absorbed into bigger cybersecurity and device-hardware roadmaps rather than staying a standalone software category -- a pattern worth watching for any founder building a point solution in AI security, since strategic investment often precedes an eventual acquisition rather than staying purely financial.
Alice's growth rate and the underlying adversarial-content problem it addresses are both real, but a 500% two-year growth rate off what was presumably a small base needs context Alice hasn't disclosed -- the company hasn't published absolute historical revenue figures, only the growth percentage and the near-$100 million current ARR, making it hard to independently verify how much of the growth is organic expansion within existing customers versus new-logo acquisition.
