OpenAI announced on 23 April 2024 that it has adopted Safety by Design principles to strengthen child safety protections across its products [source]. The company stated it is implementing measures aligned with frameworks developed by child safety organisations and technology industry groups.

The announcement outlined four core commitments: preventing the generation of child sexual abuse material (CSAM), blocking attempts to groom or sexualise children through AI interactions, restricting minors' access to age-inappropriate content, and preventing the platform from being used to facilitate child exploitation.

OpenAI stated it uses a combination of automated classifiers, user reporting mechanisms, and manual review processes to detect and block prohibited content. The company said it has implemented filters at both input and output stages to prevent CSAM-related prompts and responses. When violations are detected, OpenAI reports them to the National Center for Missing & Exploited Children (NCMEC) as required by US law.

The provider also disclosed that ChatGPT's terms of service require users to be at least 13 years old, with users aged 13–17 requiring parental consent. OpenAI stated it is developing additional tools to help parents manage their children's use of its services.

The announcement did not provide specific metrics on detection rates, false positive rates, or the volume of reports submitted to NCMEC. OpenAI stated it is working with external safety organisations including the Internet Watch Foundation and Thorn to refine its detection systems.

The Safety by Design principles were developed by technology companies, child safety advocates, and regulatory bodies to establish industry standards for preventing online child exploitation.