Alice provides end-to-end AI trust, safety and security solutions across the AI lifecycle, protecting applications, agents and models from prompt injection, jailbreaks and model exploits.
Alice is an AI trust and safety and security company founded in 2024. It provides end-to-end solutions across the entire AI lifecycle, from build to production, protecting AI applications, agents, and models from risks such as prompt injection, jailbreaks, and model exploits. The company's technical work spans adversarial research, prompt injection detection, jailbreak prevention, model exploit detection, runtime guardrails, automated red-teaming, drift detection, machine learning classifiers, and large language model evaluation. Its stated aim is to enable organizations to advance unafraid in the generative AI era.
The company's product suite, marketed as WonderSuite, comprises several components. WonderBuild handles pre-launch stress-testing of AI models; WonderFence provides dynamic runtime guardrails for AI applications; and WonderCheck delivers ongoing automated red-teaming and drift detection. Underpinning these is Rabbit Hole, an intelligence engine built on a decade of real-world adversarial research which processes billions of toxic data samples to power classifiers and safety evaluations.
Alice reports that its technology is used by 8 of the 10 leading frontier AI labs and protects over 3 billion users, securing some 1 billion daily AI-human interactions across more than half of the world's online experiences. The platform supports more than 120 languages and is deployed across financial services, healthcare, child-facing platforms, insurance, and AI/ML development. The company operates globally.
