San Francisco-based Confident Security has emerged from stealth, announcing $4.2 million in seed funding. Decibel led the investment round, with participation from South Park Commons, Ex Ante, and Swyx. The company focuses on addressing critical `AI data privacy` concerns.
## Advancing AI Data Protection
Confident Security developed CONFSEC, an end-to-end encryption tool. This solution wraps around foundational AI models. It guarantees that prompts and metadata remain private. Consequently, data cannot be stored, viewed, or used for AI training by model providers or third parties. This technology aims to remove the trade-off between using AI tools and maintaining privacy.
CONFSEC's framework draws inspiration from Apple's Private Cloud Compute (PCC) architecture. The system anonymizes data by encrypting and routing it through services like Cloudflare or Fastly. Furthermore, it employs advanced encryption, allowing decryption only under strict conditions. The software running AI inference is publicly logged for expert review, ensuring verifiable guarantees.
The company positions itself as an intermediary vendor for AI providers and their clients. Target clients include hyperscalers, governments, and enterprises. This `enterprise AI solutions` tool is also suitable for new AI browsers, such as Perplexity. Major AI companies like `OpenAI` and `Anthropic` could also offer CONFSEC to their enterprise clients. This approach helps unlock markets hesitant due to privacy concerns. Confident Security aims to enable secure AI adoption in highly regulated sectors like healthcare and finance.
