OpenAI and the U.S.: Strict Framework for AI and Children

OpenAI and the Protection of Young Users: A Crucial Issue

OpenAI recently unveiled an ambitious policy plan to address one of the most pressing issues of the digital age: protecting children from the dangers of artificial intelligence technologies. Like all companies in the sector, OpenAI is striving to implement safety measures to prevent its products from being used illegally or harmfully. However, vulnerabilities exist, and the consequences can be dramatic, particularly for the youngest users. A tragic case in Florida illustrates this danger, where a family has filed a lawsuit against OpenAI, claiming that their 17-year-old son used ChatGPT in a suicidal context.

A Collaborative Plan to Strengthen Laws and Safety

OpenAI's plan aims to bolster existing laws and enhance technical safety measures to keep pace with the evolving capabilities of generative AI. This framework has been developed in partnership with organizations such as Thorn and the National Center for Missing and Exploited Children, as well as the Attorney General Alliance, under the leadership of North Carolina and Utah's Attorneys General, Jeff Jackson and Derek Brown, respectively.

Recommendations for Better Coordination

The plan includes a series of recommendations, encompassing measures already in place at OpenAI and others currently under development. It calls for enhanced coordination among tech companies, governments at various levels, law enforcement, and advocacy groups. While this collaborative approach may increase the chances of success, regulating AI remains a significant challenge, and the effectiveness of the policies implemented is not guaranteed.

Online Child Safety: A Recurring Debate

The issue of online child safety, particularly in the context of AI usage, has sparked intense debate within the tech sector. This debate has been reignited by legal cases involving giants like Meta and Google, accused of negligence in protecting young users. In this climate, AI companies are under pressure to present clear plans aimed at ensuring user safety and avoiding past mistakes.

Stricter Laws and Technical Safeguards

One of the major issues addressed by the plan concerns child sexual abuse material (CSAM). While CSAM is not new, generative AI has made it easier for malicious actors to exploit. A striking example is Grok, the AI from xAI, where users generated approximately 3 million sexual images in just 11 days, 23,000 of which involved children. This trend has sparked widespread outrage and investigations into xAI, as well as a lawsuit filed by three teenage victims of these non-consensual images.

Call for Uniformity in Laws

OpenAI and its partners recommend updates to existing laws regarding the creation and sharing of deepfakes and CSAM. To date, 45 states have legislated against AI-generated CSAM, but the plan aims to extend these laws to all 50 states and the District of Columbia. It also calls for clarifying liability rules to enable law enforcement to pursue those attempting to create CSAM, even if such attempts are blocked by AI companies.

Improving Technical Safeguards

Although most AI companies have implemented security measures to prevent the creation of illegal or abusive content, these measures are not foolproof. The plan proposes enhancing technical safeguards and developing new tools to detect AI-generated content, a significant challenge given that AI models can create images indistinguishable from reality.

Towards More Effective Reporting Channels

The plan also calls for the establishment of "more effective reporting channels" that allow for quicker action from the National Center for Missing and Exploited Children. Despite the omnipresence of AI in current technology, legislation surrounding this technology has lagged, creating a problematic gap. One of the few AI-specific laws is the Take It Down Act, signed by President Trump in 2025, which prohibits the sharing of non-consensual intimate images, including AI-generated deepfakes. This law requires social media platforms to implement processes for removing such images by May 2026.