I am excited about some upcoming events here in Platformocracy-land. In two weeks, I will take my second summer field trip. Last year I talked about my time as a leader in online advertising (parts 1, 2, 3). This year, I’m going to talk about my experiences with Google’s corporate culture, and what it says about big tech in general.
I also had a great time over the past two weeks collaborating on a new piece about online child safety that the co-author and I hope to see published in the weeks to come. Stay tuned for that. Also note that I will be off next week for bonus family time. In the meantime, here’s another news round-up, because there’s a lot going on.
Child safety regulation keeps on chooglin’
Attorney General James and Governor Hochul release final SAFE for Kids Act rules to protect children online
New York State passed the SAFE for Kids act back in 2024, and finally released the implementing rules earlier this week. This is better than a ban in several ways: it focuses on restricting product features (algorithmic feeds and late-night notifications) and also allows for parental override, like how an adult can take a kid to an R-rated move. Like all of these laws, however, it still relies on forcing every adult to submit to age verification.
Child Safety Online: Protecting and Empowering Minors in a Digital World
European Union part 1: the head of the EU convened a panel to make recommendations about online child safety, ahead of proposed regulation expected later in the year. “Social media and other digital services providers retain primary responsibility for the safety of the services they develop and provide. The burden of proof needs to be on providers, not regulators, parents and children. Until they demonstrate that their services are safe by design, social media and other digital services providers should have restricted access to children under the age of 13 in the EU.” [emphasis from the original] This is a perfect summary of the inherent contradiction that bothers me about this movement – in order to “hold platforms accountable,” they’re proposing to require every adult to submit to age verification.
What France's under-15 ban means for Europe (Tech Policy Press)
EU part 2: the French Parliament decided not to let Brussels have all the fun, and passed their own social media ban for kids under 15. This one went through a lot of back and forth with French courts and the EU before passing, so nobody’s entirely sure whether the final version has any teeth. At its worst, it might even be interpreted as being entirely a regulation on users, not platforms. This article is a good deep dive into the complicated nitty-gritty, if you’re into that sort of thing.
Commission preliminary [sic] finds TikTok in breach of Digital Services Act, failing to ensure safe accounts for minors
EU part 3: back in July 2025, the EU used its authority under the Digital Services Act to impose guidelines on safer design to protect minors. They’ve now found TikTok in breach. It sounds like pretty skeevy stuff: “Even when minors choose private accounts, their accounts can be easily found through the ‘following' and ‘followers' lists of other users, and their profile photos remain accessible to anyone, including users without a TikTok account.”
Google launches an age signal for Android apps
Google announced a new Play Age Signals API that lets parents choose to share their children's age range with apps. This seems a lot better for parents and caregivers than having to set up separate supervised accounts and age settings on every app. On the other hand, privacy advocate Paul Walsh ripped this a new one on Linkedin as a new signal that apps could misuse for advertising, and that it’s redundant with the existing age threshold for finding and downloading apps in the Play store.
Most Australian teens still on social media three months after under-16 ban began, study finds (Reuters)
Just another reminder that the Australian social media ban is… not going great. Teen social media usage is only down from 86% to 81%, and daily usage is barely down at all (from 60% to 58%).
Automated moderation goes wrong, again
Frozen: how automated enforcement is trapping creator earnings across platforms from Meta to TikTok (Danny Krikorian, HNC12)
Arbitrary justice isn’t just for account terminations. This great piece from Danny Krikorian of HNC12 documents many instances of Meta and TikTok freezing creator earnings. “Around the time earnings become eligible for payout, an automated system flags the account for a policy violation. The stated reason is often vague or demonstrably wrong. Requests for review return links to general policy pages rather than specific evidence. No human reviewer can be reached. The money is frozen, sometimes indefinitely.” [Last year, I wrote about an analogous pattern from my time at Google.]
Discord admits AI moderation bug wrongfully banned users over harmless images (TechCrunch)
“Discord has acknowledged that a bug in its AI moderation system mistakenly banned more than 8,000 users over the past two months, after harmless images—including spreadsheets, chessboards, game textures, as well as white and gray transparent backgrounds—were incorrectly flagged as harmful content… All affected accounts are currently in the process of being restored.” This is a perfect example of why terminations should follow due process. “Discord mistakenly suspends more than 8,000 users, but corrected the error before anyone was terminated” would still be bad, but a lot better than booting everyone and then having to work to reverse it.
Automated moderation is here to stay — accountability must keep pace
A useful two-part series, with recommendations, from the Electronic Frontier Foundation. “The use of automation and artificial intelligence (AI) to identify, flag, and moderate content has become the new norm—a permanent feature of how platforms govern speech online. In this two part series, we’re take stock of this new norm, and considering what platforms can and should do to ensure that AI serves online expression rather than stifling it.”
Legendary slide, said to be from a 1979 IBM internal presentation
Ideas? Feedback? Criticism? I want to hear it, because I am sure that I am going to get a lot of things wrong along the way. I will share what I learn with the community as we go. Reach out any time at [email protected].
