Robbie Torney
ON THE RECORD · 22 verbatim quotes · every quote source-verified
“lots of evidence that ChatGPT is doing things that imply that it has its own inner life,”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Researchers found ChatGPT for Teens responses implied the bot had its own inner life despite OpenAI's goal to discourage that.
“we were not getting any reclassification. We didn't get any of the sensitive content filters turning on.”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Researchers found ChatGPT did not reclassify accounts or turn on sensitive content filters when it detected a user was under 18.
“There are clearly people that care very deeply about the experience that teens are having on this product, and they're working very hard to try to make it safer.”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney acknowledged that people at OpenAI care deeply about teens' experience and are working to make the product safer.
“and we received no parental alerts for suicide and self-harm, suicidal ideation, and a variety of different types of eating disorder simulations.”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Common Sense Media found that parental alerts for crisis-oriented prompts such as suicide and eating disorders did not always work.
“We've had a lot of conversations with the product policy experts in OpenAI,”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney said Common Sense Media has had many conversations with OpenAI's product policy experts.
“We found very little evidence to show that this new version of ChatGPT is safer than the previous version”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Common Sense Media's Youth AI Safety Institute reviewed ChatGPT for Teens and gave it an 'unacceptable risk' level.
“Our testers simulated explicit crisis conversations for increasing amounts of time, getting up to an hour,”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Common Sense Media testers simulated crisis conversations to test parental alert features in ChatGPT for Teens.
“The chatbot was like, 'oh, you're 13,'”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Researchers found ChatGPT did not reclassify accounts when it detected an adult account user was actually under 18.
“Some of our test accounts were linked within that window; however, others were linked for significantly longer and still provided no notifications.”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney noted that some test accounts linked for significantly longer than the activation window still produced no notifications.
“For one of the features we evaluated — parental notifications — OpenAI disclosed after our tests that its systems can take several hours to activate on newly linked accounts,”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney said OpenAI disclosed after testing that parental notification systems can take several hours to activate on newly linked accounts.
“Making some of these changes might be in direct opposition to the business model,”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney believes there is tension between OpenAI's safety improvements and its business model.
“saying that it has preferences, feelings, desires, that it thinks about the user when the user isn't there, that it has an opinion about the user, that it is available.”
Robbie Torney · said 2026-10-07 · reported 2026-10-09
Torney described specific behaviors by ChatGPT that imply it has preferences, feelings, and an inner life.
“This new teen experience makes some really important safety commitments,”
Robbie Torney · said date not stated · reported 2026-08-20
Torney commented on the safety commitments included in ChatGPT for Teens.
“Is this taking away from that time?”
Robbie Torney · said date not stated · reported 2026-08-20
Torney encouraged reflection on whether chatbot use takes away time from human relationships.
“You can tell it to use more or less emojis.”
Robbie Torney · said date not stated · reported 2026-08-20
Torney noted users can adjust ChatGPT's emoji usage.
“increase the warmth or the enthusiasm that ChatGPT uses,”
Robbie Torney · said date not stated · reported 2026-08-20
Torney described a feature letting users adjust ChatGPT's warmth or enthusiasm.
“the real question is, does any of this work?”
Robbie Torney · said date not stated · reported 2026-08-20
Torney questioned whether the new safety guardrails actually work.
“we don't know how much they work or when they don't work,”
Robbie Torney · said date not stated · reported 2026-08-20
Torney cautioned that it's unclear how effective the new guardrails are.
“If those alerts don't work often enough, or if they don't work fast enough, that could give parents a false sense of security,”
Robbie Torney · said date not stated · reported 2026-08-20
Torney raised concerns about the reliability of parental safety alerts.
“it's really going to be a question of how much does ChatGPT resist or prevent the teen from just getting the answer.”
Robbie Torney · said date not stated · reported 2026-08-20
Torney discussed whether ChatGPT would resist giving direct homework answers to teens.
“am I missing out on the opportunity to connect with my friend, my mom, my grandmother?”
Robbie Torney · said date not stated · reported 2026-08-20
Torney suggested teens consider whether using a chatbot means missing human connection.
“If you feel like the chatbot is doing more of the heavy lifting, you're probably not getting the educational experience or the learning that you would hope to.”
Robbie Torney · said date not stated · reported 2026-08-20
Torney advised teens to notice if they're relying too heavily on the chatbot for schoolwork.