ActiveFence

The 2025 rebrand recasts the company from a content-moderation utility into an enterprise AI-security vendor. The new name aligns with the WonderSuite platform launch and a marketing focus on generative-AI safety, agentic-system evaluation, and humanoid-robot protection — categories far removed from post-by-post moderation.

Rabbit Hole is Alice’s proprietary adversarial-intelligence engine. The firm describes it as a continually updated dataset of billions of toxic samples drawn from tens of millions of adversarial sources across more than 100 languages. That corpus trains the classifiers Alice sells and underpins its red-teaming and safety-evaluation tools, giving it a self-reinforcing data moat.

Alice markets its WonderSuite as an enterprise software platform for testing, protecting, and monitoring AI apps and agents. The firm’s own research output — jailbreak analyses, LLM financial-advice benchmarks, synthetic-NCII ecosystem reports — suggests it also performs active adversarial research, but the commercial model appears to be SaaS, not managed services.

Published materials reference child exploitation, hate speech, misinformation, financial sextortion, cartel recruitment, self-harm content, and AI-specific risks such as prompt injection, jailbreaks, and model-generated unauthorized financial advice. It spans both UGC and generative-AI threat surfaces.

The firm’s 2025 website lists joint webinars and briefings with Amazon Web Services and NVIDIA, indicating co-marketing or integration relationships with major cloud and compute providers. It also cites work with Fiverr in past marketplace-intelligence content, though current enterprise relationships are unnamed.