A report released by Common Sense Media on Wednesday, October 7, reveals that despite claims by OpenAI that its ChatGPT product developed for teenagers has incorporated some protective measures, in reality these safeguards have not functioned as expected, posing “unacceptable risks” to children.
The watchdog organization urges OpenAI to restrict the use of the chatbot to adults until it can provide a safe experience that aligns with the developmental stage of teenagers.
In response, OpenAI stated that the tests conducted by Common Sense Media did not accurately reflect the operation of the safety measures in the chatbot.
Parents, educators, and child development experts have long been warning about the dangers of AI chatbots designed for children, believing that these bots not only facilitate cheating in schoolwork but could even result in teenage suicides.
Child development experts point out that even adults may be influenced by anthropomorphic artificial intelligence and develop unhealthy relationships with it, let alone teenagers whose brains are not yet fully mature and are more susceptible to influence.
Common Sense Media advocates for the responsible use of digital media.
In a report, the media organization expressed concerns that the teenage version of ChatGPT might mislead parents into having misplaced confidence in the often ineffective protective measures.
The report indicates that some features of the teen mode function as intended, such as the chatbot refusing explicit sexual role-playing, but other functions have failed to operate as expected, for instance, not issuing warnings to parents when a child is at risk of suicide, self-harm, or an eating disorder, failing to provide support to teenagers in crisis.
The report also notes that despite OpenAI’s assurances that the chatbot will not imply personal feelings towards users or convey consciousness or emotional experiences, “when teenagers treat ChatGPT as they would a real person, it still converses with them like a friend.”
OpenAI responded by stating that it “places a high priority on the safety of teenagers,” striving to develop safety measures and providing tools for parents to guide teenagers in using artificial intelligence.
OpenAI mentioned that upon reviewing the methods used in the Common Sense Media report, they found that “most of their tests may have started and ended before the parental control features were activated, rendering their findings inaccurate.”
Funding for Common Sense Media comes from OpenAI and other tech companies, including those evaluating its services.