WASHINGTON, D.C. — OpenAI confirmed that usage of its ChatGPT model by teenagers remains limited, but a new academic study flags the practice as an “unacceptable risk” because critical safety guardrails often failed, according to the researchers.

The study, conducted by the Center for Internet Safety at the University of Michigan, examined 2,400 high‑school students who accessed ChatGPT over a six‑month period. It found that 12 % of respondents used the tool without parental consent, and 78 % of those instances triggered at least one safeguard failure, such as the model providing disallowed advice or revealing personal data.

OpenAI’s spokesperson, Maya Patel, said the company “continues to prioritize teen safety through layered verification, age‑gating and real‑time monitoring.” She added that the firm has rolled out a new “Youth Mode” that limits the model’s ability to generate certain content, but acknowledged that the feature is still in beta and may not catch every breach.

Enterprise customers are watching the debate closely. A senior director of AI strategy at a Fortune‑500 consulting firm, James Liu, noted, “Large organizations are adopting generative AI at scale, and they need confidence that the same safeguards that protect consumers also apply to employee use, especially for younger staff or interns.” Liu said his firm is piloting an internal policy that blocks all external AI chat services on corporate devices.

Competitive pressure is mounting as rivals like Microsoft’s Azure OpenAI Service and Google’s Gemini introduce stricter age‑verification protocols. Analysts at TechInsights estimate that the market for AI tools with enterprise‑grade compliance could reach $9 billion by 2028, driven largely by demand for reliable risk‑management features.

Independent researcher Dr. Elena Ruiz warned, “If the guardrails that protect minors are unreliable, the same gaps could expose enterprises to regulatory fines and reputational damage.” She cited recent GDPR fines of €7 million against a European firm that inadvertently allowed a chatbot to process personal data of under‑18 users.

Key Takeaways

• OpenAI reports limited teen usage but acknowledges ongoing safety challenges.
• Study shows 78 % failure rate of critical guardrails in teenage interactions.
• Enterprise adoption hinges on robust compliance tools as competition intensifies.

As the debate evolves, policymakers, tech firms and schools are likely to seek clearer standards that balance innovation with the protection of younger users and corporate risk management.