Anthropic is committed to implementing robust child safety measures in its generative AI technologies, alongside other leading companies, as part of a new initiative led by Thorn and All Tech Is Human to prevent the misuse of AI to create or spread child sexual abuse material. The organization has implemented rigorous policies, conducts red teaming, and collaborates with external experts to ensure model safety, strictly prohibiting content that describes or supports child sexual exploitation. Anthropic is also working towards specific mitigations, including responsibly sourcing training data, detecting and removing CSAM, conducting stress testing, defining policies, and prohibiting customer use of models for further harm against children. The company aims to detect abusive content, include user reporting options, incorporate prevention messaging, and maintain the quality of its mitigations.