Common Sense Media unveiled an AI ratings framework in 2023

The design

Common Sense Media, a nonprofit focused on kids and technology, first announced plans for an AI ratings and reviews system on 17 July 2023, then revealed the system's framework on 15 September 2023, naming OpenAI and Khan Academy as its first participating companies. Both dates and details come from the organization's own July 2023 press release and its September 2023 framework announcement, retrieved via the Wayback Machine after the live pages were reorganized under the site's current AI-ratings section. The framework describes the goal as a 'nutrition label' for each AI product, scored against eight named Common Sense AI Principles.

What the evidence says

The organization's own releases cite a poll it commissioned with Impact Research, reporting that 82% of surveyed parents wanted a rating system for AI products and only 40% knew of a reliable source of information on AI's suitability for children; this is the organization's account of its own commissioned poll, not an independently audited survey, and this entry treats it as such. The framework document lists the eight principles by name, including 'Help People Connect,' which asks whether a product 'fosters meaningful human contact' rather than displacing it, a principle directly relevant to companion-style AI, though the September 2023 announcement does not name any specific companion app.

What it asks of people

The framework asks AI developers to submit products for review against principles including privacy protection, fairness, transparency and child safety, and asks families and policymakers to treat the resulting ratings as an independent, third-party signal rather than a company's own claim about its product. Founder James Steyer frames the project as a response to what he calls social media's earlier regulatory failure, a comparison that is the organization's own stated rationale rather than a finding from the ratings process itself.

Privacy and safeguards

Among the eight named principles, 'Defend Privacy' asks whether a rated product 'protects data' and provides 'appropriate transparency and control,' and 'Keep Kids and Teens Safe' asks whether it protects safety and well-being regardless of intended audience; the September 2023 announcement states these as evaluation criteria for future ratings, not as findings about any product rated so far, since the framework announcement predates specific product ratings.

• Which specific AI companion products has Common Sense Media since rated under this framework, and how did they score on 'Help People Connect'?
• Does the 82% figure from the commissioned poll appear in any independently conducted survey of the same question?
• How has the framework's original eight-principle structure changed between its 2023 launch and the platform as it exists today?

The 2023 announcements fix a starting point: a named framework, a named set of principles, and two initial industry participants. What they do not yet show is how consistently that framework was applied once ratings for AI companion products specifically began to appear.