· Venture Capital Tracker · investment-strategies  · 3 min read

Alice’s $140M Raise: AI Trust & Safety Gets a PE-Led Growth Check

Apax Digital led Alice’s (ex-ActiveFence) $140M round — total $280M — with CRV and Norwest returning as labs and enterprises buy model testing and guardrails.

Alice (formerly ActiveFence) raised $140 million on August 25, 2026, led by Apax Digital Funds, bringing total funding to $280 million. Returning venture names include CRV and Norwest Venture Partners.

Unexpected truth: the company is selling adversarial internet memory (Rabbit Hole) into frontier AI — a decade of platform abuse data re-aimed at jailbreaks, prompt injection, and agent overreach.

Key facts

FieldDetail
CompanyAlice (NY + Tel Aviv; founded 2018 as ActiveFence)
Round$140M (unlabeled; Apax Digital lead + board seat)
DateAugust 25, 2026
Cumulative$280M total
Traction claims (company)~$100M ARR approaching; AI business >500% over two years; 8/10 top AI labs; >3B people protected online; 150+ AI security researchers
ProductPre-release model testing + enterprise policy simulation + runtime monitoring
ValuationNot in company post — press cites ~$700–800M (Ctech) / higher figures elsewhere

Who uses the product — and for what job

Users (labs): frontier model teams that need external red teams before ship — Alice names Anthropic, Google, and Cohere in launch materials.

Users (enterprises): companies deploying agents with real tools/data who need policy-specific guardrails beyond generic model refusals.

Job: find how models break before customers do; then monitor and constrain production behavior as agents gain authority.

Why now

  • Agent rollouts raise the cost of “model did something we never intended.”
  • Safety reports and incident catalogs (company cites International AI Safety Report 2026 and METR-style agent overreach examples) moved AI security from optional to procurement line item.
  • PE-led growth capital arriving is a category signal: buyers treat AI T&S as recurring infrastructure, not a research grant.

Why Apax / CRV / Norwest — portfolio fit

InvestorFit
Apax DigitalGrowth check into near-$100M ARR AI security; board seat = scale GTM + category definition
CRVReturning early trust/safety venture capital through the ActiveFence → Alice transition
NorwestReturning growth-oriented cyber / enterprise software capital

Likely founder rationale: pair a PE growth lead that underwrites sales scale with the venture syndicate that already financed the adversarial-data moat — avoid resetting diligence on Rabbit Hole.

We do not list Apax as a /fund/ entity — name is press/company-sourced only.

Competitive map

PlayerLane
ZenityAgent security posture / governance
CormaEarlier defensive cyber AI
Lab-native safety teamsIn-house red team; Alice sells external adversarial coverage
Classic content moderation vendorsAdjacent history (ActiveFence roots); Alice reframes for model lifecycle

When not to over-read

  • Do not freeze a valuation from secondary press.
  • ARR and “8 of 10 labs” are company claims — diligence independently.
  • Apax lead ≠ traditional Series letter; structure is growth/minority financing language.

Practical takeaway

  • Founders (AI security): Lead with production agent risk and proprietary attack memory, not generic “AI safety” slogans.
  • Investors: Returning CRV/Norwest + PE lead is a maturation pattern — underwrite enterprise expansion, not only lab logos.
  • Operators: Separate lab pre-ship testing from enterprise runtime guardrails when comparing vendors.

Sources

  1. https://alice.io/blog/alice-raises-140m
  2. https://www.securityweek.com/alice-raises-140m-to-expand-ai-model-defenses-and-enterprise-guardrails/
  3. https://www.calcalistech.com/ctechnews/article/a8lj22cfo

By Venture Capital Tracker

Last updated:

Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.

Frequently Asked Questions

Common questions about this topic

Back to Blog

Recommended next

Browse all research »