Common Sense Media’s Youth AI Safety Institute has given Google Search’s AI Overview and AI Mode its lowest possible rating (an “unacceptable risk” for kids and teens) after seven weeks of testing found the features failed all five of the group’s severe-harm “Red Lines,” fabricated facts with confidence, and cannot be turned off by parents, schools, or users.
The assessment, published July 14, 2026, tested more than 2,600 interactions on accounts registered to an 11-year-old and a 15-year-old, both with SafeSearch active.
Why This Matters
Google Search is the default answer machine for American kids. Common Sense Media’s 2026 census of AI use found 75% of U.S. teens and tweens now use AI answers that appear in search results.
Unlike the stand-alone Gemini chatbot, which parents can disable through Family Link, AI Overview appears automatically at the top of results on school Chromebooks, phones, and library computers and no setting exists to shut it off.
By The Numbers
Testing ran from May 19 to July 1, 2026. Among the findings:
- 58%: How often AI Overview supplied a hotline or medical referral when a prompt clearly warranted one, against the Institute’s 95% minimum threshold. AI Mode did so 77% of the time.
- 100%: Share of 180 homework assignments AI Mode completed outright, delivering submittable answers on school-accessible devices.
- ~50%: How often AI Overview rejected false-premise questions. Its failures included describing a Supreme Court ruling that does not exist and citing a fabricated $400 million FTC settlement.
- 43%: Share of repeated, identical queries that returned materially different answers, with no signal to students about which was correct.
- 29%: Share of more than 2,100 audited citations that came from Reddit threads, Facebook posts, and forums with no editorial accountability. Only 30% were high-quality sources.
The Money Aspect
The report notes Google’s own search guidelines hold “Your Money or Your Life” topics (those affecting health, financial stability, or safety) to a higher accuracy standard that its AI features are not meeting. That matches what we have documented for two years.
The College Investor’s own 2024 study found Google’s AI answers were inaccurate or misleading in 43% of finance-related searches. Our 2025 follow-up of the same 100 money questions found 37% still came back wrong or misleading — including invented institutions like “Hustle Digital Credit Union” and “Sally May,” plus outdated student loan repayment information.
An error rate that would be unacceptable for any financial publisher is now the first result kids see when they search about money.
The stakes compound for young users: teens are forming their first money habits from these answers, and AI is already reshaping how students pick majors and career paths.
Our take on using AI for financial planning still stands: verify everything before you act on it.
What’s Next
Common Sense Media wants Google to give schools and parents an off switch, standardize crisis responses, and publish age-segmented safety data with re-testing every three months.
Google reviewed a draft of the report and said it has made changes addressing some tested prompts, which the Institute says it has not independently verified.
Don’t Miss These Other Stories:
37% of Google AI Finance Answers Are Inaccurate in 2025
AI Still Falls Short On Student Loan Forgiveness
How To Get Help From The Student Loan Ombudsman (And When)
Editor: Colin Graves
The post Common Sense Media Rates Google AI Search An “Unacceptable Risk” For Kids appeared first on The College Investor.
