Leaked chats, fake doctors, and 98 students who wrote an AI rulebook
For Educators · The Social Institute
AI is showing up in students’ private conversations, their health feeds, and now their policy proposals. Here’s what happened and what it means for your classroom.
- AI chats aren’t as private as students think: Shared conversations with the AI chatbot Claude turned up in Google and Bing search results this week, with Bing showing roughly 612 results at one point, a reminder that hitting “share” makes a chat public.
- Fake AI “doctors” are going viral on TikTok: Research from the digital marketing agency Hallam found AI-generated content in 40% of top health videos, 84% of “health tips” results, and top fake-doctor videos averaging about 2.5 million views each.
- Students wrote their own AI policy: 98 teens from all 50 states drafted and passed a model “Students First Act” for AI in K-12 schools by a vote of 82-16, and AASA plans to share it with 10,000 school leaders.
Students think their chats with AI are private. A Google search shows otherwise
Shared chats with the AI chatbot Claude showed up in Google and Bing results, a reminder that privacy is a skill, not a setting.
The Gist
Over one weekend, people found that some seemingly private conversations with the AI chatbot Claude appeared in Google and Bing search results. The chats were exposed when users had clicked “share” to send someone a link, and search engines then indexed those pages. What feels private with an AI tool isn’t always private, especially once you share it. We know that a scare and restrict approach doesn’t work with students, but we can empower and equip them to build the habit of pausing before they share. Students who protect their privacy like they’re famous stay in control of their own information.
What to Know
According to Wired, some shared Claude conversations became searchable online, including people asking which political party to join and other questions they assumed were private. Claude lets users create a public link to share a brief look at a chat. Anthropic had instructed search engines not to include those pages in results, but the instruction didn’t work the way it’s supposed to. At one point, Bing showed about 612 results for shared Claude chats.
The lesson reaches past any one chatbot. As Anthropic pointed out, once you share a conversation, you’re making that content public, like anything else you post online. These AI tools can feel like a trusted confidant, but your content can be saved or seen by people you never thought it would reach. For educators, this is a good moment to talk with students about what they share with AI and where that information can end up. Students who understand that online privacy rarely means true confidentiality are better prepared to make smart choices.
TSI’s Take
While many platforms have strong privacy settings that a user can choose to use, protecting your privacy is still a skill students need to build. When students learn to think before they share, they stay in control no matter which app they’re using.
Educators can help students protect their privacy like they’re famous by:
- Talking about what “private” really means online: Help students see that a shared link or a saved chat can travel far beyond who they meant to reach.
- Encouraging a pause before sharing: Before hitting share or typing something personal, ask, “Would I be okay with this being public?” Remember, a screenshot of a private chat lasts forever.
- Reviewing privacy settings together: Walk through where the share and privacy controls are in the apps and AI tools students actually use.
Help students take charge of their personal information with the #WinAtSocial Lesson, AI and Data Privacy, where students learn about the benefits and risks of sharing personal data.
Fake AI “doctors” are going viral on TikTok, and the advice is often completely wrong
AI-generated content showed up in 40% of top health videos on TikTok, and the fake doctors behind them are averaging about 2.5 million views each.
The Gist
New research found that fake, AI-generated “doctors” are spreading questionable health advice on TikTok and pulling in millions of views. These accounts are created to look like real physicians, and they’ve posted everything from disproven cancer myths to made-up miracle products. AI-generated content appeared in 40% of top health videos, so students are scrolling past a lot of untrue yet convincing advice. The good news? Students can learn to spot these fakes and find sources they can trust.
What to Know
According to The Guardian, research from the digital marketing agency Hallam found that:
- AI-generated content appeared in 40% of top health-related TikTok videos
- For searches like “health tips,” 84% of videos were AI-generated or AI-assisted
- The top AI “doctor” videos averaged about 2.5 million views each
Some of the myths these fake doctors promote? Microwaving food in plastic or sleeping next to your phone causes cancer. Others promoted odd remedies and even a made-up anti-aging product no one could find for sale. A health misinformation researcher called it “industrialised exploitation of trust,” because the accounts borrow the look and authority of a real doctor to earn views and interactions.
Looking legit and being legit are two different things. With about 1 in 5 people turning to social media for health information, students need to be able to tell a credible source from a convincing fake.
TSI’s Take
With 57% of students getting their news from social media, the goal isn’t for students to distrust everything online. It’s to equip them with the critical thinking skills to determine what’s real so that they can make smart, informed choices. Educators can help students find credible influencers and think critically by:
- Teaching students to check the source: Before believing or sharing health advice, ask who’s behind it and whether a credible expert or organization backs it up.
- Naming the signs of AI-generated content: Talk about the tells, like a too-perfect delivery or claims that sound too good to be true.
- Building the habit of a second look: Encourage students to cross-check a surprising claim with a trusted source before they act on it or pass it along.
Help students separate fact from fiction with the #WinAtSocial Lesson, AI and Misinformation, where students learn to spot misleading content and check what’s real.
Student Voice for the Win: With Congress stalling on AI in schools, 98 students took it into their own hands
98 teens from all 50 states passed their own model AI policy 82-16, and 10,000 school leaders are about to see it.
The Gist
Over a single July weekend, 98 teens from all 50 states gathered in a replica of the U.S. Senate chamber and drafted and passed a model policy for AI in K-12 schools, taking their learning into their own hands. Their “Students First Act” passed 82-16 and acknowledged the exact questions schools are wrestling with, like when students should use AI and how to prevent it from replacing real learning. Students have serious, authentic ideas about the tech in their lives when we give them a seat at the table (or when they create their own table), and schools don’t have to wait for a law to start that conversation.
What to Know
According to NPR, the students gathered at the Edward M. Kennedy Institute in Boston, hosted by groups including Day of AI and AASA (The School Superintendents Association). With no national policy for AI in schools, they built their own from dozens of proposals.
Their bill took some clear stances: teach AI literacy as soon as students get classroom devices, ban AI on graded tests, and let teachers ask a student to defend their work out loud if they suspect AI misuse. A big worry behind their choices was that classmates might lean on these tools to do their thinking for them. As one student put it, some kids will “use it as a brain” rather than a tool. Another said, “We missed the mark with social media and phones,” and they stressed that it should not be a repeated mistake.
Students aren’t asking to use AI with no limits. They want guidance, and they have strong opinions about what responsible use looks like. AASA plans to share the students’ bill with 10,000 school leaders who are currently writing their own AI rules.
TSI’s Take
When students help create the rules, they’ll understand them more clearly and be more likely to buy-in. The schools getting AI right are the ones bringing students into the conversation. Educators can build on this by:
- Co-creating AI expectations with students: Huddle with your class at the start of the year to define when AI helps learning and when it goes too far.
- Asking students what responsible use looks like: They’re already forming opinions, so give them a place to share and shape them.
- Connecting AI choices to their goals: Help students see how using AI as a tool, not a shortcut, protects the learning and skills that prepare them for their futures.
Help students find their voice in shaping how AI shows up at school with the #WinAtSocial Lesson, School Tech Policies: Devices, AI, and Beyond, where students explore what responsible AI use looks like and build the habits that will serve them all year.
A chat that wasn’t as private as it looked, a doctor who doesn’t exist, and 98 students who decided to write the rules themselves. Students make better AI choices when they’ve had the chance to practice making them. That’s exactly what #WinAtSocial is built for. Want to see how it looks in your school? Request a demo and we’ll walk you through it.