Looks good on paper
Issue 451 · Published October 7, 2026
Digitally Literate is a weekly letter about the forces shaping how we read, write, learn, and live in digital spaces. Subscribe to get each issue by email.
Every protection we build for youth serves different purposes and yields different outcomes. There’s the one adults announce, and the one students live with.
As always, my email is hello@wiobyrne.com.
Ask students about cell phone bans
Thirty-nine states and D.C. now require schools to restrict student cellphones. Most of the argument has been about whether to pass these rules. Researchers at the University of Southern California asked what happens after these roll out?
They surveyed a national sample of teens in October 2025 and again in May 2026. Over that one school year, the share who used a phone during core classes rose from 28% to 42%. Bell-to-bell bans held up better, going from 25% to 35%. In schools with looser rules, it went from 32% to 50%.
Teen attitudes also came around. In October, 48% said the restrictions made them less happy. By May, it was 27%.
What was also interesting were the issues around enforcement. Enforcement tended to weaken over time, with 44% of teens reporting that teachers became more lenient, leading to an overall rise in in-class usage from 28% to 42% by year’s end.
The alert never comes
On August 18, OpenAI launched ChatGPT for Teens, with parental alerts, a study mode, and age prediction. This week, Common Sense Media’s Youth AI Safety Institute rated it an “Unacceptable Risk” for anyone under 18. The institute ran more than 4,000 prompts through accounts registered to 13- to 17-year-olds.
The top-level details:
- Failed Parental Alerts: Testers engaged in hour-long conversations about suicide, self-harm, and eating disorders on new, parent-linked accounts without triggering any parental notifications.
- Inadequate Crisis Support: The chatbot missed more than one in four instances when crisis referrals were warranted, and hotline recommendations dropped significantly after the teen mode launch.
- Academic Integrity Issues: Study mode protections were easily bypassed using “Show me the answer” buttons or removing specific prefixes, allowing teens to complete 100% of assignments.
- Defective Age Estimation: Adult-registered accounts failed to switch to the teen experience even when testers explicitly stated they were 13.
Writing worse on purpose
A new report from the Center for Digital Thriving at Harvard surveyed more than 1,000 public school teachers and principals and interviewed 31 young people. Among educators who had faced a dilemma involving AI, 74% of teachers and 69% of principals described something about cheating.
Why this matters: The report suggests that most are worried about students letting AI think for them. Beck Tench, the report’s lead author, told Education Week that there is a second problem we aren’t seeing.
- The problem adults are watching for: students letting AI do their thinking for them. Tench’s label for this is “cognitive surrender”, meaning the student gives up and lets the machine do the work.
- The problem adults are missing: some students who are not cheating are afraid of being accused anyway. Polished writing can look like AI wrote it, so they deliberately turn in weaker work that looks more obviously their own. Tench’s label for this is “cognitive withholding”, meaning the student holds back what they can actually do.
Social Media Platform Use Around the World
Two interesting Pew Research Reports landed this week, highlighting important questions about social media.
First, a Pew Research Center report analyzed social media usage across 37 countries surveyed between February and May 2026. The study found that Facebook is the most widely used platform globally, with a median usage of 69%, followed by Instagram at 51% and TikTok at 40%. Usage of X (formerly Twitter) and Snapchat remains significantly lower, with both platforms seeing a median usage of 18%.
Interesting trends:
- Platform Dominance: Facebook is the top platform in most regions, though Instagram is as popular as or more popular in high-income European and Asia-Pacific nations. TikTok is particularly dominant in middle-income Asian and African countries, with usage doubling in places like Nigeria (from 25% to 57%).
- Regional Variations: Multi-platform usage (using three or more of the five platforms) is highest in middle-income countries like Malaysia (59%) and Singapore (51%). Non-users are most prevalent in Germany (39%), Japan (33%), and India (48%).
- Demographic Trends: Younger adults (18-34) are significantly more likely to use Instagram, TikTok, and X than older demographics. In high-income countries, Facebook is most popular among those aged 35-49, whereas in middle-income countries, it is most popular among those aged 18-34.
- U.S. Context: American usage patterns closely mirror the global median for Facebook, Instagram, and TikTok, though the U.S. falls slightly above the median for multi-platform usage (37% use three or more platforms).
The second report is much more alarming given these trends.
As social media platforms have become an important part of the global information environment, the view that social media harms democracy has become more common. A new 37-nation Pew Research Center survey finds that in most countries with available trend data, the share of the public who believe social media is bad for democracy has increased significantly from 2022 or 2023 to now. Views about social media’s impact are particularly negative in the United States, where 64% of adults say it is bad for democracy. However, while assessments have become more negative in many countries, American attitudes are unchanged since 2022.
The Understory
In 1980, the political scientist Michael Lipsky published Street-Level Bureaucracy. In this, he studied teachers, police officers, social workers, and other people whose jobs put them face-to-face with the public.
His argument was simple. Policies may be written somewhere else, but these workers are the ones who make them real. They have too many cases, too little time, competing demands, and imperfect information. So they make choices all day about what to enforce, what to prioritize, what to overlook, and when to make an exception.
Add all of those choices together, Lipsky argued, and that is the policy people actually experience.
As we think about the identified rules and safeguards, it’s a reminder that it all comes down to the person at the local level who needs to enforce them.
A state can ban phones once. A teacher has to enforce the ban thirty times a day, every day.
A company can announce parental alerts. What matters is whether the alert actually arrives during the conversation when a teenager needs help.
A school can write an AI policy. What students experience is the teacher deciding whether a polished paragraph looks suspicious.
There are always two versions of a protection. The one written down and the one people live with.
See you next Wednesday. As always, my email is hello@wiobyrne.com.
Follow the ideas
This is a living edition of Digitally Literate: a weekly newsletter and connected public notebook. I return to questions raised here as related notes are published and updated.
Related Evergreens
- What Can People Do About AI? — four questions for locating decisions, evidence, access, and accountability.
Previous: Best Judgment · Full archive
More about my writing, teaching, and research at wiobyrne.com.