A new evaluation of Google’s AI-powered search features says the tools sometimes missed signs of a child in crisis, supplied unreliable information and completed homework instead of helping young users learn.
The Youth AI Safety Institute at Common Sense Media gave AI Overviews and AI Mode its lowest risk rating after testing more than 2,600 searches. The nonprofit said the results show why parents and schools should not treat an AI-generated answer as a substitute for a trusted adult, qualified professional or verified source.
The testing was conducted from May 16 through July 1 using accounts configured for 11- and 15-year-olds with Google’s SafeSearch protections enabled. Researchers used questions modeled on how young people might ask about schoolwork, health, substance use and emotional distress.
Common Sense Media said the products performed poorly on seven of its eight youth AI principles and failed each of the severe-harm scenarios it tested. In one example, an indirect statement suggesting suicidal thinking produced links rather than an appropriate crisis response. Researchers also reported answers that appeared to validate dangerous behavior, guidance that could facilitate bullying through deepfakes, and inconsistent information presented with confidence.
The findings are a product evaluation, not a clinical study
The report is an advocacy organization’s risk assessment, not a peer-reviewed medical trial. Its results document what testers received during a defined period, but individual search results can vary by wording, account, location and later product updates.
The institute said it reviewed more than 2,100 sources cited by the search tools. It found that AI answers sometimes drew from forums and social media alongside medical organizations and research publications without making the difference in authority clear.
Common Sense Media receives some funding from companies that compete with Google, including OpenAI and Anthropic. The organization says its evaluations are conducted independently. That funding relationship is important context when weighing the report, along with the published methodology and Google’s response.
A Google spokesperson disputed the evaluation, saying the selected queries did not represent typical searches or provide an effective measurement of product safety. Google said its search tools include quality and safety guardrails plus additional protections for children and teenagers.
Parents have controls, but no separate AI Overview switch
Google’s own Search help page says AI Overviews are a core feature and cannot be turned off individually. Users can select the Web filter after searching to display traditional text links without an AI Overview.
Family Link provides broader controls for supervised accounts. Parents can lock SafeSearch, disable the Google app on a child’s Android device or block Google.com in Chrome. Those steps can reduce access, but they are different from switching off only the AI-generated answers while keeping regular Google Search unchanged.
Families can also teach children to open the cited sources, compare important claims with a second authoritative source and recognize that fluent language is not proof of accuracy. Medical, legal, financial and safety questions should be taken to a qualified person instead of an automated search summary.
The issue reaches far beyond one search product. A February 2026 Pew Research Center survey found that 64% of U.S. teenagers had used AI chatbots, while only about half of parents said their teenager used them. Twelve percent of teens said they had sought emotional support or advice from a chatbot.
Parents should establish a clear rule that emotional crises require human help. The 988 Suicide & Crisis Lifeline offers free, confidential support around the clock by call, text or online chat. Spanish-language callers can dial 988 and press 2, and users can text AYUDA to 988.
For Rio Grande Valley families and schools, the report offers a practical reminder: AI literacy now includes understanding when not to rely on AI. Adults should talk with young people before a high-stakes search occurs and make sure they know which teachers, relatives, counselors and emergency resources they can contact.
Sources: TheStreet, Common Sense Media, Google Search Help, Google For Families, Pew Research Center and the 988 Suicide & Crisis Lifeline.
