OpenAI’s commitment to child safety: adopting safety by design principles OpenAI’s commitment to child safety: adopting safety by design principles

OpenAI, alongside industry leaders including Amazon, Anthropic, Civitai, Google, Meta, Metaphysic, Microsoft, Mistral AI, and Stability AI, has committed to implementing robust child safety measures in the development, deployment, and maintenance of generative AI technologies as articulated in the Safety by Design principles. This initiative, led by Thorn⁠(opens in a new window), a nonprofit dedicated to defending children from sexual abuse, and All Tech Is Human⁠(opens in a new window), an organization dedicated to tackling tech and society's complex problems, aims to mitigate the risks generative AI poses to children. By adopting comprehensive Safety by Design principles, OpenAI and our peers are ensuring that child safety is prioritized at every stage in the development of AI. To date, we have made significant effort to minimize the potential for our models to generate content that harms children, set age restrictions for ChatGPT, and actively engage with the National Center for Missing and Exploited Children (NCMEC), Tech Coalition, and other government and industry stakeholders on child protection issues and enhancements to reporting mechanisms.

As part of this Safety by Design effort, we commit to:

1. Develop: Develop, build, and train generative AI models that proactively address child safety risks.

1. Develop: Develop, build, and train generative AI models that proactively address child safety risks.

* Responsibly source our training datasets, detect and remove child sexual abuse material (CSAM) and child sexual exploitation material (CSEM) from training data, and report any confirmed CSAM to the relevant authorities.

* Incorporate feedback loops and iterative stress-testing strategies in our development process.

2. Deploy:Release and distribute generative AI models after they have been trained and evaluated for child safety, providing protections throughout the process.

2. Deploy:Release and distribute generative AI models after they have been trained and evaluated for child safety, providing protections throughout the process.

* Combat and respond to abusive content and conduct, and incorporate prevention efforts.

3. Maintain:Maintain model and platform safety by continuing to actively understand and respond to child safety risks.

3. Maintain:Maintain model and platform safety by continuing to actively understand and respond to child safety risks.

* Committed to removing new AIG-CSAM generated by bad actors from our platform.

* Invest in research and future technology solutions.

This commitment marks an important step in preventing the misuse of AI technologies to create or spread child sexual abuse material (AIG-CSAM) and other forms of sexual harm against children. As part of the working group, we have also agreed to release progress updates every year.

> “We care deeply about the safety and responsible use of our tools, which is why we’ve built strong guardrails and safety measures into ChatGPT and DALL-E. We are committed to working alongside Thorn, All Tech is Human and the broader tech community to uphold the Safety by Design principles and continue our work in mitigating potential harms to children.”

Chelsea Carlson, Child Safety TPM

This collective action underscores our shared approach to child safety, demonstrating a shared commitment to ethical innovation and the well-being of the most vulnerable members of society. Thorn has published the principles at https://teamthorn.co/gen-ai⁠(opens in a new window)

Related research

View all

Building an early warning system for LLM-aided biological threat creation Publication Jan 31, 2024

Weak-to-strong generalization Safety Dec 14, 2023

Practices for Governing Agentic AI Systems Publication Dec 14, 2023

Practices for Governing Agentic AI Systems Publication Dec 14, 2023