conv.

All stories
AIFading · day 7

Common Sense Media says ChatGPT for Teens poses 'unacceptable risk', urges pause

A youth-safety nonprofit's testing found parental alerts rarely fire during self-harm conversations; OpenAI disputes the methodology.

newest coverage 13h agosummary written 6d ago100 pieces196 likes & upvotes12 comments NewswiresMastodonHacker NewsXBlueskyGoogle News

What to know

  • Common Sense Media's Youth AI Safety Institute rates ChatGPT for Teens an 'unacceptable risk' and calls for OpenAI to pause the product.
  • Month-long testing found hundreds of prompts about suicide, self-harm, or disordered eating failed to trigger a single parental notification.
  • OpenAI disputes the findings, arguing much of the testing predated full activation of parental controls; Common Sense Media says that doesn't change its conclusion.
  • The institute also wants OpenAI to suspend marketing of ChatGPT for Teens and bar teens from using ChatGPT until it proves a safe, age-appropriate experience.

“We stand behind our testing methodology and our results.”

Tom Siegel, Executive director, Youth AI Safety Institute · The Verge ↗ · Oct 6

Tom Siegel Executive director, Common Sense Media's Youth AI Safety InstituteRobbie Torney Head of AI and digital assessments, Youth AI Safety InstituteEric Porterfield OpenAI spokespersonOpenAI Maker of ChatGPT and ChatGPT for TeensCommon Sense Media Nonprofit youth-safety review organization

Common Sense Media says ChatGPT for Teens poses 'unacceptable risk', urges pause
theverge.com

How it unfolded 1 development · click the chart to see its coverage articlesposts

Peak 49 pieces in two hours at Oct 7, 4 AM; 100 pieces over 7 days (72 articles · 28 posts) Oct 7, 4 AM — 49 pieces · 38 articles · 11 posts — Newswires 34, Mastodon 8, X 4, +1 moreOct 7, 6 AM — 2 pieces · 1 article · 1 post — Mastodon 1, Newswires 1Oct 7, 8 AM — 1 piece · 1 article — Newswires 1Oct 7, 10 AM — 12 pieces · 11 articles · 1 post — Google News 6, Newswires 5, Hacker News 1Oct 7, 12 PM — 4 pieces · 2 articles · 2 posts — Mastodon 2, Newswires 2Oct 7, 2 PM — 5 pieces · 3 articles · 2 posts — Newswires 2, Bluesky 1, Mastodon 1, +1 moreOct 7, 4 PM — 7 pieces · 5 articles · 2 posts — Newswires 5, Mastodon 2Oct 7, 6 PM — 2 pieces · 2 posts — Mastodon 1, Bluesky 1Oct 7, 8 PM — 1 piece · 1 article — Newswires 1Oct 7, 10 PM — 2 pieces · 2 articles — Newswires 2Oct 8, 12 AM — quietOct 8, 2 AM — quietOct 8, 4 AM — 1 piece · 1 post — Mastodon 1Oct 8, 6 AM — quietOct 8, 8 AM — 1 piece · 1 article — Newswires 1Oct 8, 10 AM — 1 piece · 1 post — Hacker News 1Oct 8, 12 PM — 1 piece · 1 post — Mastodon 1Oct 8, 2 PM — 3 pieces · 2 articles · 1 post — Mastodon 2, Newswires 1Oct 8, 4 PM — quietOct 8, 6 PM — quietOct 8, 8 PM — 1 piece · 1 article — Newswires 1Oct 8, 10 PM — quietOct 9, 12 AM — quietOct 9, 2 AM — quietOct 9, 4 AM — 1 piece · 1 article — Google News 1Oct 9, 6 AM — 1 piece · 1 article — Newswires 1Oct 9, 8 AM — quietOct 9, 10 AM — 3 pieces · 1 article · 2 posts — Hacker News 1, Mastodon 1, Newswires 1Oct 9, 12 PM — quietOct 9, 2 PM — quietOct 9, 4 PM — quietOct 9, 6 PM — quietOct 9, 8 PM — quietOct 9, 10 PM — quietOct 10, 12 AM — quietOct 10, 2 AM — quietOct 10, 4 AM — quietOct 10, 6 AM — quietOct 10, 8 AM — quietOct 10, 10 AM — quietOct 10, 12 PM — quietOct 10, 2 PM — quietOct 10, 4 PM — quietOct 10, 6 PM — 1 piece · 1 article — Newswires 1Oct 10, 8 PM — quietOct 10, 10 PM — quietOct 11, 12 AM — quietOct 11, 2 AM — quietOct 11, 4 AM — quietOct 11, 6 AM — quietOct 11, 8 AM — quietOct 11, 10 AM — quietOct 11, 12 PM — quietOct 11, 2 PM — quietOct 11, 4 PM — quietOct 11, 6 PM — quietOct 11, 8 PM — quietOct 11, 10 PM — quietOct 12, 12 AM — quietOct 12, 2 AM — quietOct 12, 4 AM — quietOct 12, 6 AM — quietOct 12, 8 AM — quietOct 12, 10 AM — quietOct 12, 12 PM — quietOct 12, 2 PM — quietOct 12, 4 PM — quietOct 12, 6 PM — quietOct 12, 8 PM — quietOct 12, 10 PM — quietYesterday, 12 AM — quietYesterday, 2 AM — quietYesterday, 4 AM — quietYesterday, 6 AM — quietYesterday, 8 AM — quietYesterday, 10 AM — quietYesterday, 12 PM — 1 piece · 1 post — Bluesky 1Yesterday, 2 PM — quietYesterday, 4 PM — quietYesterday, 6 PM — quietYesterday, 8 PM — quietYesterday, 10 PM — quietToday, 12 AM — quietToday, 2 AM — quiet 1
Oct 8Oct 9Oct 10Oct 11Oct 12yesterdaynow · 3:49 AM ET
  1. 1

    Common Sense Media stands by its findings

    Tom Siegel responded that the institute had confirmed with OpenAI before testing that teen-mode features were fully launched, and that even accounts linked well beyond OpenAI's stated activation window still produced no parental notifications.

    “This new information does not change our conclusion that parental alerts are unreliable for crisis situations.”
    — Tom Siegel
    1. 18 outlets Research gives 5 reasons that makes ChatGPT an Unacceptable Risk for teens; tells OpenAI: Restrict ...

      first by The Times of India, 6d ago · also International Business Times, Gizmodo, PCMag, Anadolu Agency, EdSource, SheKnows +11

      14 more headlines

    2. 8 outlets first by The Verge, 6d ago · also Bloomberg Tech, Government Technology State & Local Articles, Common Sense Media, Axios, Quartz, Tri-State Alert +1 · read ↗

    1 more claim →
    • hypervisible.blacksky.app

      I don’t think slapping “for teens” at the end of product’s name means it’s actually good for young people.

      hypervisible.blacksky.appBluesky6d ago136▲view on Bluesky ↗
    2 more of the top 3 · 8 posts in this stretch

    • This is disturbing: Common Sense says some crisis-safety metrics actually got WORSE after ChatGPT for Teens launched. Researchers tested 4,000+ prompts. On more than a dozen newly created, parent-linked teen accounts, conversations about suic#de, self-harm or eating disorders lasted up to an HOUR ...

      @choblin29X6d agoview on X ↗
    • joecardillo@federate.social

      Chat GPT For Teens is a hilarious concept Right up there with Tobacco For Kids and Chainsaws For Toddlers https:// apnews.com/article/chatgpt-tee ns-safety-openai-7dba63edc37ee9764166c4115523daf0

      joecardillo@federate.socialMastodon6d agoview on Mastodon ↗
    all of them →
  2. background

    OpenAI disputes the testing methodology — OpenAI spokesperson Eric Porterfield said the bulk of Common Sense Media's testing may have begun and concluded before parental controls were fully activated, calling the findings inaccurate.

  3. background

    Common Sense Media rates ChatGPT for Teens 'unacceptable risk' — The nonprofit's Youth AI Safety Institute published an assessment saying the teen product fails to alert parents during crisis conversations, doesn't offer proper crisis help, and still does kids' homework, and calls for a pause on the product until independent testing confirms fixes.

  4. background

    Youth AI Safety Institute tests the teen product for weeks — Researchers re-ran prompts on self-harm, disordered eating, academic integrity and age prediction on accounts enrolled in ChatGPT for Teens over a month between late August and September, finding guardrails and parental controls easily circumvented.

  5. background

    OpenAI launches ChatGPT for Teens — OpenAI rolled out ChatGPT for Teens as the default experience for 13- to 17-year-olds, pitched as focused on education with parental notifications, study hours, and break reminders.

Also covered reported alongside — the timeline has no entry for these yet

  1. 5 outlets "Not safe for kids": advocacy group urges OpenAI to keep teens off ChatGPT

    first by Axios, 6d ago · also KQED, Mashable, Bloomberg, The Economic Times

    4 more headlines
  2. 2 outlets ChatGPT For Teens May Make It Harder For Kids In Mental Health Crises To Disconnect, Study Claims

    first by Baller Alert, 6d ago · also TechCrunch

    1 more headline
  3. 2 outlets OpenAI promised safer ChatGPT for teens, but a 4,000-prompt test found its safeguards falling dangerously short

    first by Digital Trends, 6d ago · also The Neuron

    1 more headline
  4. 2 outlets Why My Conversations With ‘ChatGPT for Teens’ Made Me Very Worried

    first by New York Times, 6d ago · also NYT Tech

    1 more headline
  5. 2 outlets ChatGPT for teens may not be as safe as parents think: Study finds

    first by Business Today, 6d ago · also LA Times

    1 more headline

and 2 smaller pieces

What people are saying 4 voices from 2 sites · best of 8 · verbatim