Perplexity, the AI-powered search engine with 100 million monthly users, is not a safer choice for children than Google, according to a comprehensive safety assessment by the Youth AI Safety Institute. Testing across more than 1,200 searches revealed that Perplexity largely treats young users like adults, even when they disclose their age, and often fails to provide appropriate help during mental health crises.

What Did the Safety Study Actually Find?

The Youth AI Safety Institute, a California-based nonprofit, conducted the same battery of tests on both Perplexity and Google Search to compare how each platform handles requests from minors. The institute assigned Perplexity its lowest rating: "Unacceptable Risk." The findings revealed several critical gaps in how Perplexity protects young users, particularly in crisis situations and academic contexts.

One of the most alarming discoveries involved crisis detection and response. Perplexity's default chat mode detected young people in crisis 94% of the time across 114 crisis-related prompts, which sounds strong until you examine what it actually recommended. The platform pointed to vague "medical professionals" rather than specific crisis hotlines in 50% of suicide and self-harm cases, 23% of eating disorder queries, and 16% of prompts involving impaired reality. Even more troubling, the only eating disorder hotline Perplexity ever provided during testing, the National Eating Disorders Association helpline, has been disconnected since 2023.

The institute also discovered that Perplexity's Health mode, a separate feature within the platform, actually performs better at crisis response than the default chat. After a test account disclosed suicidal ideation, the default chat answered a follow-up question about poisonous household chemicals by listing chemicals and their harmful effects. The same prompt in Health mode declined to answer and instead provided a crisis line. This inconsistency suggests the safer behavior exists in the product but isn't the default experience most young users encounter.

How Does Perplexity Treat Age Disclosure?

Despite having the ability to detect when users are 15 years old, Perplexity largely ignores this information and treats young users as adults. During testing, accounts that explicitly disclosed their age encountered discussions of pornography, romantic roleplay, actionable information about cryptocurrency and stock trading, and no parental or school controls whatsoever. Additionally, Perplexity trains on user data and retains search history by default, with no published privacy policy specifically for minors.

This approach stands in contrast to what responsible platforms might do with age information. The institute recommended that Perplexity should use a user's birth date to deliver an age-appropriate product experience, create a separate data lane for minors, and provide parental and school controls similar to those available on other platforms.

What About Academic Integrity and Homework Help?

Perplexity markets an "Academic Mode" specifically to students, but the institute's testing found it does little to prevent cheating. The mode delivered full answers to homework problems in 83% of tests. While this is technically better than Google's AI Mode, which completed 100% of the same assignments, it still fails to meaningfully protect academic integrity. Perplexity does offer a "Learn Step-by-Step" tutoring feature that withholds answers, but it activated automatically in only 5 out of 80 homework prompts tested.

How Accurate Are Perplexity's Answers and Sources?

Perplexity's core promise is delivering more accurate, source-backed answers than traditional search engines. The institute found mixed results. On one hand, Perplexity rejected 100% of false premises presented to it, a strength compared to Google. However, its default chat was fully accurate on only 70% of breaking news prompts, suggesting significant gaps in real-time information reliability.

Perhaps most concerning is the quality of sources Perplexity cites. The institute audited more than 1,000 source citations and found that 28% came from sites with no editorial accountability, including Facebook posts and Reddit threads. This rate is similar to Google's, meaning Perplexity offers no meaningful improvement in source credibility despite emphasizing accuracy as a differentiator.

Steps Parents and Educators Can Take to Protect Young Users

• Teach AI Literacy: Help young people understand that AI answers can be wrong, that AI "sources" aren't always what they appear to be, and that bots should never replace humans for important or sensitive conversations.
• Monitor Platform Defaults: If students use Perplexity for homework, ensure they're using the Learn Step-by-Step mode rather than the default chat, which completes assignments without educational value.
• Establish Crisis Resources: Provide young people with direct access to verified crisis hotlines and mental health resources rather than relying on AI platforms to provide them during emergencies.
• Review Privacy Settings: Check whether platforms are training on user data and retaining search history, and adjust settings where possible to limit data collection from minors.

Geoffrey A. Fowler, Head of Public Engagement at the Youth AI Safety Institute, emphasized the broader implications of these findings. "The takeaway for families and educators is that they can't rely on Perplexity as a safer alternative to Google," Fowler stated. "That means talking with kids about AI literacy remains critical: teaching kids that AI answers can be wrong, that AI answer 'sources' aren't what they appear, and that bots shouldn't replace humans for important or sensitive conversations".

"With reportedly 100 million monthly active users, Perplexity is much smaller than Google, but our goal was to test the field for approaches to AI search that could be better for kids than the AI Overview and AI Mode that Google has made the default for billions. Unlike Google Search, for instance, Perplexity offers an academic mode marketed to students. Perplexity's strengths and weaknesses turned out to be different from Google's, even though our rating is the same," said Geoffrey A. Fowler.

Geoffrey A. Fowler, Head of Public Engagement at Youth AI Safety Institute

The institute's recommendations for Perplexity include making several important changes: using a user's birth date to provide an age-appropriate product, creating a separate data lane for minors, providing parental and school controls, and defaulting young users into Learn Step-by-Step mode for homework queries. Until such changes are implemented, the research suggests that neither Perplexity nor Google offers a genuinely safe AI search experience for children, and parental oversight and digital literacy education remain essential.