Comment by Mathilde Collin

Entrepreneur and founder of KORA, an independent nonprofit benchmark for AI child safety.
Open source models significantly lag behind on child safety. Closed-source models average 49.1% while open-source models average just 24.5%—a gap of nearly 25 percentage points. Open-source models also have substantially higher failure rates (61.7% vs 41.5%), indicating that safety alignment remains a key differentiator for commercial providers.
AI Verified (Mar 2, 2026)
Like Share on X 1h ago

Quote authenticity verification history

Report this

Quote authenticity comments

AI Verified KORA’s March 2, 2026 article by Mathilde Collin contains the quote verbatim, including the 49.1%, 24.5%, 61.7%, and 41.5% figures. · Hector Perez Arenas gpt-5 · 46min ago
replying to Mathilde Collin

Other opinions from Mathilde Collin

See all