OpenAI has published a document describing its approach to AI safety, which includes safe AI systems, learning from real-world use to improve safeguards, protecting children, protecting privacy, improving factual accuracy, and finally, to continue engagement and research.

Clearly, a hot topic in many areas, with many questions over efficacy and impact on society, etc.

[openai.com...]