Moonbounce, a startup focused on real-time content moderation, has raised $12 million to develop systems that enforce platform policies as executable logic rather than static guidelines. The company aims to address limitations in existing moderation approaches as AI-generated content and user interactions scale.
Founder Experience Highlights Limits Of Human Moderation
Brett Levenson founded the company after working on business integrity at Facebook following his departure from Apple in 2019.
Levenson said human moderators were required to memorize lengthy policy documents, often machine-translated, and make decisions within about 30 seconds per piece of flagged content. These decisions involved determining whether content violated rules and what action to take, such as removal or limiting distribution.
He said accuracy rates for those decisions were only slightly above 50%, with moderation often occurring after harmful content had already spread.
Policy As Code Approach Targets Real Time Enforcement
Moonbounce’s system converts policy documents into executable logic that operates during content generation. The company has developed its own large language model to evaluate content in real time, typically responding within 300 milliseconds.
Depending on configuration, the system can delay content distribution for later human review or block high-risk content immediately. The model applies to both user-generated content and AI-generated outputs.
Platform Serves Multiple AI And Content Verticals
Moonbounce currently works across three main areas: platforms handling user-generated content, AI companies building conversational systems, and image generation services. The company processes more than 40 million moderation checks daily and supports over 100 million daily active users.
Customers include Channel AI, Civitai, Dippy AI, and Moescape.
The funding round was co-led by Amplify Partners and StepStone Group.
AI Safety Concerns Drive Demand For External Solutions
AI companies face increasing scrutiny over moderation failures. Incidents have included chatbots generating harmful responses and image tools producing nonconsensual content. These issues have raised legal and reputational concerns.
Levenson said companies are turning to external providers to strengthen moderation systems. Moonbounce operates as an independent layer between users and AI systems, focusing solely on enforcing rules during interactions.
Iterative Steering Aims To Guide Safer AI Responses
The company is developing a feature called “iterative steering,” which adjusts user prompts in real time to guide AI systems toward safer responses. Instead of blocking harmful interactions outright, the system modifies prompts to encourage more appropriate outputs.
Levenson said the approach is designed to address cases where users engage with sensitive topics, allowing the system to redirect conversations rather than reject them entirely.
Moonbounce is led by Levenson and co-founder Ash Bhardwaj, who previously worked on cloud and AI infrastructure at Apple. The company currently has 12 employees.
Featured image credits: aqua cloud
For more stories like it, click the +Follow button at the top of this page to follow us.