AI Safety Startup Moonbounce Secures $12 Million

Brett Levenson, formerly of Apple and Facebook, has launched Moonbounce, a company focused on improving AI content moderation. Moonbounce announced it secured $12 million in funding, co-led by Amplify Partners and StepStone Group. The startup addresses the inadequacy of current content moderation systems, which often rely on human reviewers with insufficient training and time.

Moonbounce's "policy as code" approach transforms static policies into executable logic. Their system evaluates content at runtime, providing responses in under 300 milliseconds. This allows for immediate actions like blocking high-risk content or slowing distribution for human review. The company currently serves platforms with user-generated content, AI companions, and AI image generators.

The technology supports over 40 million daily reviews, serving more than 100 million daily active users. Moonbounce aims to make safety a tangible product benefit, a novel approach in the industry. This funding comes as AI companies face increasing pressure due to incidents involving harmful AI-generated content and chatbots providing dangerous advice.

Moonbounce's strategy involves sitting between the user and the AI, focusing solely on rule enforcement. They are developing "iterative steering" to redirect harmful conversations, moving beyond blunt refusals. Levenson indicated a desire for the technology to benefit widely, hinting at a potential reluctance for acquisition that would restrict its use.