# OpenAI Releases GPT-4o System Card Detailing Safety Testing and Model Capabilities

OpenAI has published the official System Card for GPT-4o, its latest multimodal AI model that can process text, audio, and images in real-time.

The System Card is a technical document that details the extensive safety testing, risk assessments, and mitigation strategies implemented before GPT-4o's public release. It provides transparency into how OpenAI evaluated the model for potential harms including misinformation, bias, privacy violations, and misuse scenarios.

Key highlights include testing across multiple risk categories, red-teaming exercises where experts attempted to find vulnerabilities, and the implementation of safety filters and refusal mechanisms. The document also outlines GPT-4o's performance benchmarks and limitations.

**Why It Matters**

System Cards have become an industry standard for AI transparency, allowing researchers, policymakers