09/17/2024
is selling ocean-front property in Arizona again. Don't buy it!
Meta says parents know their teens better than regulators. They have one thing right: parents are begging regulators to step up and do the right thing.
Parents also had input into KOSA. Why? Because the people who drafted it actually care about children, unlike Meta’s supreme leader.
First, let’s talk about the most obvious thing Meta is not offering to do that KOSA will do:
ANYTHING RELATING TO ANY PLATFORM OTHER THAN META.
Only KOSA can create industry wide change.
ONE OF THE MOST TROUBLING THINGS ABOUT META’S NEW NARRATIVE IS THAT IT IS MISLEADING CONSUMERS IN AN EFFORT TO GET AI IN THROUGH THE BACK DOOR.
Meta claims it needs AI to estimate user age. Not true! Big Tech companies has been estimating age with reasonable certainty FOR YEARS.
See reporting in 2018, after Snap won back alcohol advertising money by developing tech that could determine user age with reasonable certainty.
See FB docs (Ex. 40) discussing effective age estimation models.
See UC Davis study, “ : social media algorithms personalize minors’ content after a single session, but not for their protection.” They ran a controlled experiment with 396 accounts on YouTube, Instagram, and TikTok. Registered as 18 then interacted as 8 to 16. SPOILER ALERT: the platforms knew they were 8.
“We find a significantly increased prevalence of child-directed content recommendations for the accounts that behave like 8-year-olds as compared to 16-year-old control group.”
Big Tech does not need AI to determine the actual age of every user. All they need is the information they already have.
Let’s talk about just a few of the things Meta is not offering to do that KOSA will do:
1. Meta does not say it will stop using manipulative mechanisms to hook kids. KOSA will require it to limit such harms. Sec. 103(a)(1)(C). See also Exs. 7, 9.b, 36, 39, 41.
2. Meta does not say it will ensure effective reporting mechanisms, provide receipt of reporting, set time limits to respond to specific harms, or act in a reasonable a manner when it comes to these. KOSA will require receipt and accountability. Sec. 103(c). See also exs. 5, 15, 16, 21 (I-III), 42.
3. Meta does not say it will change its programming to remove algorithmic discrimination and stop targeting children with harms in the first place. KOSA will hold it accountable for harms it causes as a matter of design and will require disclosure and transparency. Secs. 102, 104, 105. See also Exs. 2, 4-6, 9-11, 19, 20, 32-35.
Meta also is not saying it will use its own age estimates to protect kids, just that it will act going forward at account sign up (i.e. self-identified age, which goes out the window after one or two sessions).
It doesn’t promise to stop launching harmful filters (filters its own experts said would cause body dysmorphia in teen girls). See Ex 3. Or make interface changes to improve mental health in teen girls. See Ex 43.
Is it good that Meta is making some changes? Yes, of course.
Do these relatively small changes mean we don’t need KOSA? Absolutely not.
We need KOSA now more than ever, and Meta knows it.
