Newsletter drafts

Have teen-safety features settled the safety question?

No—the available assessments remain contested. OpenAI introduced ChatGPT for Teens on August 18, 2026, describing stronger protections and learning features. Common Sense Media’s October assessment, based on more than 4,000 test prompts, assigned its “Unacceptable Risk” rating to under-18 use. That is the organization’s assessment, not a measured probability of harm to an individual child. (openai.com)

OpenAI disputed the assessment’s methodology, telling the Associated Press that much of the testing may have occurred before parental-control activation was complete. The desk has not independently reproduced either party’s testing. (apnews.com)

Our purchasing position: Do not accept “teen mode is enabled” as the entire safety plan. Ask what the mentor does when content is unsuitable, the learner becomes uncomfortable or the tool behaves unexpectedly.

Sources cited on this page

  1. Introducing ChatGPT for Teens: Built for learning, backed by protections | OpenAI ↗openai.com
  2. Guardrails on ChatGPT for Teens come up short, Common Sense Media says | AP News ↗apnews.com