Alice Raises $140M as AI Security ARR Approaches $100M

In an August 25, 2026 financing release, Alice disclosed a $140 million round led by Apax Digital Funds, taking cumulative funding to $280 million. MoreTech, Phoenix Financial and existing investors participated, while Apax Digital gained a board seat.

SiliconANGLE’s same-day account added Samsung, SentinelOne and Maj Invest to the investor list and placed company-reported annual recurring revenue near $100 million, with the AI portion of the business growing more than 500% over two years. Those traction figures are operating indicators supplied by Alice, not results accompanied by audited accounts or enough underlying data for independent verification.

The financing is clear; the valuation is not

The new capital equals half of Alice’s disclosed lifetime funding. That is a substantial addition to its balance sheet, but the available deal materials do not specify how much is primary capital, whether secondary share sales formed part of the transaction or what ownership the investors received.

An independent comparison of the valuation estimates records a figure close to $1 billion from one account and a range of $700 million to $800 million in Israeli coverage, while noting that Alice’s transaction release contained no valuation. The company can therefore be described as approaching unicorn territory only with that qualification; none of those figures is a formally published deal term from Alice.

Using the company-reported ARR threshold as an illustrative denominator, the lower valuation range implies a multiple of roughly seven to eight times ARR, while the higher estimate implies about ten times. The actual multiples would be somewhat higher because the recurring-revenue figure is approaching the threshold rather than already at it. This comparison is provisional, not a conventional valuation analysis based on complete financial statements.

The ARR headline is useful but incomplete

ARR is the strongest commercial signal released with the financing because it is intended to represent recurring contracted business. It is not equivalent to revenue recognized under accounting rules, cash collected or profit, however, and the public materials provide no calculation methodology, gross margin, retention rate, contract duration or customer-concentration data.

The growth percentage has a narrower scope than the headline revenue measure. It applies to the AI business rather than explicitly to total company revenue or ARR, and its starting value and measurement basis are not public. It could refer to revenue, bookings or another internal indicator, so direct comparisons with publicly traded cybersecurity vendors would be unreliable.

The distinction matters because Alice combines an established trust-and-safety operation with newer model-security products. Without a revenue split, investors and customers cannot determine how much recurring business comes from legacy platform work and how much comes from adversarial testing, runtime controls or production evaluation for AI systems.

Rabbit Hole turns trust-and-safety data into security products

Alice’s official platform description identifies the company as the business formerly known as ActiveFence and presents Rabbit Hole as a proprietary collection containing billions of adversarial examples across more than 120 languages. The scale, uniqueness and “largest dataset” positioning remain vendor assertions rather than independently benchmarked findings.

The commercial repackaging is nevertheless concrete. Patterns gathered while tracking fraud, extremism, coordinated manipulation and other online abuse become test inputs for WonderBuild, which probes models, applications and agents before deployment for weaknesses such as jailbreaks, prompt injection and data leakage.

WonderFence applies related intelligence at runtime by inspecting inputs and outputs against production guardrails. WonderCheck extends the model after launch through periodic testing for drift, regressions and emerging vulnerabilities. Together, the products let Alice sell the same underlying abuse knowledge at several points in an AI system’s lifecycle instead of treating the archive solely as content-moderation data.

This feedback model is the central technical proposition behind the financing: observed attacks supply test cases, testing informs guardrails, and production behavior can reveal new patterns for later evaluation. Its value depends on whether the archive provides relevant coverage as attack techniques change and whether the controls perform reliably in customer environments—outcomes for which Alice has not released comparative benchmarks in connection with the round.

The next evidence must come from financial disclosure

The financing gives Alice substantial resources to expand its platform, the team maintaining Rabbit Hole and its commercial organization. Apax Digital’s board representation also creates continuing investor involvement beyond the capital injection.

The unresolved questions are now financial rather than conceptual. A precise transaction valuation, audited revenue, retention data, margins and a revenue split between trust-and-safety services and AI security remain unavailable. Until those figures emerge, the round and cumulative capital are confirmed financing facts, while the ARR, growth and dataset-scale figures should remain clearly labeled as company-reported operating claims.

Also read:

• Aureka Raises $100M—Its Biological Model Still Faces Clinical Proof
• Cloud Security Testing Starts With Ownership—Scanners Come Later
• Hush Security Raises $30M to Give Every Enterprise AI Agent an Identity