Security, Safety, and Trust for the AI Era

Alice secures the AI applications and agents enterprises run in production, where a wrong model response can create real security, compliance, brand, or safety problems. The platform covers the full lifecycle in one place: it red-teams AI systems before launch, enforces guardrails at runtime, and re-tests them continuously once they’re live, all under one set of policies and a single audit trail. The platform is informed by Alice’s AI safety work, helping seven of the 10 largest foundation model labs align their models.

Alice is model-agnostic, so it protects every model and agent in your environment, with guardrails tuned to your own policies rather than generic benchmarks. Coverage is multimodal and multilingual, and runtime enforcement happens at sub-120ms latency. Compliance reporting maps to the EU AI Act, ISO 42001, NIST AI RMF, and OWASP Top 10 for LLMs.

Detection is powered by Rabbit Hole, Alice’s adversarial intelligence engine. Rabbit Hole draws on nearly a decade of real-world attack data across billions of users, so the system recognizes adversarial patterns that actually appear in production rather than ones generated in a lab.

Alice (formerly ActiveFence), founded in 2018, works with customers including Amazon, TikTok, Nvidia, and Meta and has protected more than 3 billion people. The company brings years of trust and safety expertise to enterprise AI security. Alice is based in New York and Tel Aviv and has raised $130 million from investors including CRV, Highland Europe, Grove Ventures, and Norwest Venture Partners.