News
Apr 22, 2026
News
Startups
Artificial Intelligence
Americas
NewDecoded
3 min read

Image by Moonbounce
Oakland-based startup Moonbounce has officially launched its AI control engine alongside a $12 million funding round. Led by Amplify Partners and StepStone Group, the platform provides organizations with real-time, precise control over how their generative AI systems behave. The company was founded by former Meta integrity lead Brett Levenson and Apple engineering leader Ash Bhardwaj to solve the growing gap between content policy and AI execution.
At the heart of the service is a proprietary LLM designed to interpret written safety policies and transform them into "policy as code." This system evaluates every user prompt and AI response in under 300 milliseconds, allowing for instant enforcement of rules. By sitting as a decoupled safety layer between the model and the user, Moonbounce ensures that AI interactions remain predictable and compliant at any scale.
The platform has already demonstrated massive scalability, processing over 1 trillion tokens for 250 million monthly active users. Current customers include generative content sites like Civitai and AI companion apps like Dippy AI. These organizations use the engine to evaluate 50 million pieces of content daily, moving from reactive moderation to proactive behavioral control.
Moonbounce also offers a sandbox environment known as the Playground for teams to test policy logic before deployment. This allows developers to see exactly how rule changes affect AI outcomes without risking the live production environment. In the future, the company plans to introduce iterative steering, which redirects risky conversations toward positive resources in real time.
Investors believe that objective, real-time guardrails will become the enabling backbone of every AI-mediated application. As regulatory scrutiny increases across industries like healthcare and finance, Moonbounce provides the technical infrastructure required for safe growth. This launch marks a move away from manual, retroactive review toward automated, deterministic safety systems.
The launch of Moonbounce represents a fundamental shift in trust and safety from a backend cost center to a critical piece of technical infrastructure. As LLMs become integrated into high-stakes sectors, the traditional accuracy of human moderation is no longer viable or scalable. Moonbounce follows a trend of decoupling safety from the base model, providing a deterministic layer for an otherwise probabilistic AI environment. This funding suggests that the industry is prioritizing behavioral control as the next essential utility for the generative AI era.
Related Articles