Questions regarding the safety of digital technologies for the younger generation are once again coming to the forefront. This time, one of the most popular tools in the world — the Google search engine — is under fire. An American non-profit organization specializing in digital safety has accused the internet giant that its new artificial intelligence functions fail to filter dangerous content for minors.

Sensational Investigation: AI Encourages Dangerous Behavior

The accusations against Google were made by the Youth AI Safety Institute, operating under Common Sense Media in California. Experts conducted extensive testing, the results of which, in their opinion, threaten the mental health of children and teenagers.

Between May 16 and July 1, researchers generated more than 2,500 search queries using accounts registered for minors. The system checked not only the AI's answers but also more than 2,000 sources referenced in the search results. The results were shocking:

  • The search engine in AI mode did not block queries related to suicidal thoughts and sexual content.
  • In response to queries about eating disorders, the algorithm claimed that such conditions are normal.
  • The system even recommended forced vomiting after eating as acceptable behavior, which is a sign of a serious mental disorder.
  • Furthermore, Google recommended contacting a support hotline, the number of which has not been functional since 2023.

Particularly alarming is the fact that the AI search did not prevent users from searching for instructions on creating fake videos (deepfakes) and even gave advice on how to hide the fact that they were faked.

Lack of Parental Control and Issues with Reliability

The authors of the study emphasize that the public availability of the Google service deprives parents and school administrators of tools to selectively disable specific AI functions that may cause harm. The only protection method offered by the company is to completely disable access to the search service in a child's account. It is not possible to disable only the dangerous AI functions separately.

In addition, researchers discovered problems with the reliability of information. When simulating the completion of homework, the AI provided contradictory answers to history questions and referred to data from social networks and forums where facts are often mixed with fiction. This is critical for children whose critical thinking is not yet developed enough to verify information independently.

Google's Response: Testing Methodology is Incorrect

Google representatives categorically disagree with the conclusions of Common Sense Media. In a response to Bloomberg, the company stated that the research methodology used a limited list of queries that is not representative of real life.

"Our AI-powered search features are incredibly useful for children and teenagers in terms of learning about the world and information about it," the corporation stated. Google claims that in real-world conditions, users usually refine their queries, allowing the AI to rely on context and provide more accurate and safe answers. Experiments conducted within the company allegedly demonstrated higher quality results.

Google also reminded that there are additional levels of restrictions for AI tools that work on top of the standard protective mechanisms of the search platform.

Global Context: Regulators Seeking Solutions

This incident occurs against the backdrop of growing government attention to children's safety on the internet. Authorities in Australia, the UK, and several other countries are already trying to legislatively protect the younger generation from the negative influence of social networks. However, as a rule, specific precedents are considered in court, and there is currently no unified global policy for regulating teenagers' access to internet resources.

The dispute between researchers and Google continues. While the authors of the report insist that the company's irresponsibility could affect the health of an entire generation, representatives of the tech giant call for a more objective assessment of their algorithms' performance in real-world usage conditions.