Alice
Detection, threat intelligence and model red teaming for platforms and AI labs, and formerly ActiveFence.
Also trades as ActiveFence, Spectrum Labs
About Alice
Alice is an Israeli-American artificial intelligence safety company. It assists technology companies to track content such as disinformation, hate speech, fraud, and child abuse, across user-generated content platforms and generative AI systems. WonderSuite is the AI security platform for teams launching customer-facing AI in regulated industries.
Founded in 2018 by Noam Schwartz, Alon Porat, Eyal Dykan and Iftach Orr, the company is co-headquartered in New York City and Tel Aviv, Israel. It was known as ActiveFence until it rebranded to Alice in January 2026. By mid-2021 it had about 200 employees across six offices, up from roughly 60. Named clients include NVIDIA, Amazon, TikTok and Cohere. Alice describes itself as keeping over 3 billion people protected.
WonderBuild red teams customer-facing AI apps, agents and agentic workflows before launch. WonderCheck runs ongoing red teaming in production. WonderFence provides runtime guardrails tuned to a customer's application and policies.
Drafted from en.wikipedia.org · not yet confirmed by Alice
Open jobs
From Alice's own job board.
NewAlice · M4 · Ramat Gan
NewAlice · P2 · Ramat Gan
Alice · P3 · London
Showing 5 of 9
Locations
- Companies here
- 5 jobs
- 2 jobsCompanies here
- 1 jobCompanies here
What they offer
In Tools & Services. Listed, not recommended.
Recently
From Alice's own newsroom feed and from the hiring team.
The Alice Labs Cyber RL Benchmark v1.0: Can AI replace a security analyst?From their feed
We tested 10 frontier models on 12 real cyber tasks with planted decoys and objective scoring. Who did the best job?
October 6, 2026
Alice Guardrails in LiteLLM: Screen Every Call, LiveFrom their feed
Alice is now a guardrail provider in LiteLLM (v1.101.0). Turn on runtime screening for prompts, responses, and tool calls in minutes, with no SDK, no code.
October 6, 2026
ENT-IPI Bench: Indirect Prompt Injection BenchmarkFrom their feed
We evaluated eight frontier models as enterprise agents across 147 adversarial scenarios for their vulnerability to indirect prompt injection.
October 6, 2026
From Policy Document to Inline Classifier in 24 HoursFrom their feed
Turn a policy PDF into a gated, versioned classifier scoring 100% of traffic inline at under 120ms p95. See how the pipeline handles week three.
October 6, 2026
What a Former Google Cloud CISO Says About AIFrom their feed
Former Google Cloud CISO Phil Venables on the AI security risk leaders keep missing, why fundamentals still do the heavy lifting, and how to think about securing agents.
October 6, 2026
The AI Risk Debt Your Enterprise Is Already Carrying | AliceFrom their feed
Most enterprises have guardrails. Few have visibility into what their AI is actually doing. Alison Cossette of ClariTrace breaks down the risk debt quietly building in agentic systems and the three things leaders need to implement now.
October 1, 2026