Helix Security raises $42M to launch SecureFrame, privacy-preserving ML model auditing

AI · 4 min read

Helix Security announced a $42 million Series A and launched SecureFrame, a platform designed to audit machine learning models with a focus on privacy-preserving evaluation. SecureFrame provides tools for black-box and white-box testing, membership-inference resistance checks, and certified audit reports that can be presented to partners or regulators without exposing sensitive training data.

The platform leverages cryptographic techniques like secure multi-party computation and zero-knowledge proofs to validate model properties while keeping datasets confidential. SecureFrame is positioned for industries such as healthcare, finance, and telecom where model performance needs to be demonstrable to auditors but underlying data cannot be shared.

Helix will use the financing to expand compliance automation features, develop industry-specific audit templates, and grow partnerships with assurance firms. Beta users include a fintech startup and a medical imaging provider preparing for external audits.