August 10 Tech and Democracy Roundup: KOSA’s Big Week, and America’s Two Playbooks for Protecting Kids Online.
It’s Monday, August 10 — welcome back to your Tech & Democracy roundup.
Let’s kick things off with 3 big stories:
The White House finalized its voluntary AI oversight framework, but hasn’t released it publicly. The framework exempts open-weight models entirely, and companies are reportedly encouraged to hold off sharing models until they’re as close to public release as possible. Five Senate Democrats demanded visibility into the framework, and the Foundation for American Innovation filed FOIA requests to force the administration to make it public.
Illinois passed a landmark social media safety law (more on this below). The Children’s Social Media Act, set to take effect in 2028, will force social media sites to limit addictive design features, including prohibiting sending notifications to minors between 10pm and 7am. It’s the second major tech bill Governor JB Pritzker signed this summer.
Kentucky Gov. Andy Beshear signed an executive order setting new standards for data centers in the state. Developers must submit an energy plan showing they won’t drive up utility bills for residents. The order also gives the state discretion to deny permits for any project judged to negatively affect air quality and water supply.
Now, for our main story, let’s unpack recent developments in the effort to protect kids on social media.
It’s no secret that teen mental health has worsened over the past 15 years. Diagnosed anxiety among all U.S. teens is up 61% since 2016, and more than half of teen girls now report persistent sadness, up from about a third in 2011. Jonathan Haidt has been the most prominent evangelist for the case that social media is the primary culprit for the shift. Others caution against asserting a causal link between social media and declining teen mental health, pointing to factors that rose alongside it. What is clear — from study after study after study — is that excessive screen time and social media use is linked to harm, especially for kids. Lawmakers worldwide have taken note.
Outside the U.S., outright bans are picking up steam. In December, Australia became the first major country to ban under-16s from major platforms like TikTok, Instagram, Facebook and YouTube. France followed in July, and the UK and EU look poised to do the same.
The jury is out on whether bans actually work. A study published this month by Australia’s own internet safety regulator found more than eight in ten Australian teens are still using major social media sites months after the ban took effect.
But it may be a moot point here in the U.S. “There’s no magical birthday when you earn your First Amendment rights,” Jenna Leventoff, Senior Policy Counsel at the ACLU, told The Renovator, adding that minors share largely the same speech and access rights as adults.
Instead, state lawmakers have settled on two broad approaches to social media reform: the “parental consent” model (popular in red states), and the “platform behavior” model (popular in blue states).
Utah pioneered the former approach, passing the nation’s first app store age verification and parental consent law last year. Utah’s law, which inspired similar measures in Texas, Louisiana, and Alabama, doesn’t target Meta, X, or TikTok directly. Instead, it places the burden on Apple and Google to verify a user’s age when they attempt to download any app and to obtain a parent’s consent before letting a minor proceed. Several other Republican-led states are considering similar bills.
The second approach targets platform behavior. Rather than gating access, laws like the one Illinois just passed place the burden on social media companies to implement design changes to limit features that harm minors, like infinite scrolling, algorithmic feeds, and dark patterns that keep users hooked. California, Maryland, New York, and Colorado have passed similar laws.
Each approach has champions on Capitol Hill. Sen. Mike Lee’s (R-UT) App Store Accountability Act (ASAA) mirrors the red-state model but remains stuck in committee and is highly unlikely to pass this session. The Kids Online Safety Act (KOSA), the popular but contested Senate bill that would impose a federal “duty of care” on social media platforms, is much further along. The Senate Commerce Committee voted unanimously last week to advance it, teeing up a full Senate floor vote.
If you’re thinking ‘haven’t we been here before with KOSA’, it’s because we’ve been here before with KOSA. It’s been re-introduced in the Senate every session for the past four years, and was passed in 2024 (in a 91-3 vote) before dying in the House. As described previously in The Renovator, the two chambers remain at odds over the “duty of care” language. House lawmakers worry it’s vague and exposes tech companies to frivolous lawsuits, while Senate supporters say any bill without this provision lacks teeth.
KOSA has fierce advocates and critics in civil society. Child-safety groups like Fairplay and Common Sense Media have backed KOSA since its introduction, and Issue One’s Council for Responsible Social Media (which Danielle Allen sits on) has made its passage a top priority. “This is a major step toward finally giving families the protections they deserve online,” said Craig Obey, SVP of Federal Affairs at the mental health advocacy group Inseparable, in a statement after last week’s vote.
But privacy and civil-liberties groups like the ACLU and Electronic Frontier Foundation oppose KOSA. “Sponsors say the law is written to regulate design features, but when you dig into it, it still limits the content kids can see online,” Jenna told The Renovator. “Look at how the bill defines design features — things like notifications, or filters that change how your face looks. A lot of that is really just content.” The ACLU raises similar objections to age-verification proposals, arguing the data collection they require creates surveillance risks (they support comprehensive privacy legislation and digital literacy education instead).
The bottom line: states aren’t waiting on Congress — red states have largely pursued age verification, while blue states have focused on platform design. In Washington, KOSA has more momentum than any kids’ safety bill in years. Whether the House and Senate can resolve their “duty of care” standoff is the key question moving forward.
Here’s What Else is Happening
On the subject of kids’ safety, Courts have had plenty to say of late:
A New Mexico judge ordered Meta to pay $567 million for creating a ‘public nuisance’ by fueling the state’s youth mental health crisis. The order, which comes after a March verdict ordered the company to pay $375 million, also bars push notifications to minors between 10pm and 7am (akin to the law Illinois just passed).
A federal judge ruled Meta, Google, and TikTok can’t block California’s social media law (the one mentioned earlier) from restricting algorithmic “For You” feeds for minors.
Federal Agencies
The FTC announced it will stop enforcing ‘disparate impact’ claims — a legal theory regulators have used to challenge algorithmic bias and discrimination – following a Trump executive order calling for the doctrine’s elimination.
The FCC banned foreign-made humanoid robots from being sold in the U.S., citing national security risks and the need to shield American robotics firms from Chinese competition.
Congress
Sens. Ted Cruz (R-Texas) and Raphael Warnock (D-Ga.) introduced a bill to create a bipartisan federal commission advising Congress on the ethical and human-dignity implications of AI.
Sen. Ron Wyden (D-Ore.) unveiled a draft proposal to end existing tax incentives for data centers and create a new excise tax on their construction. The revenue would be directed to workers and communities disrupted by the AI buildout.
Reps. Greg Casar (D-Texas), Valerie Foushee (D-N.C.) and Sara Jacobs (D-Calif.) introduced a bill that would tax large AI companies to fund jobs for workers displaced by automation.
States
It was a mixed bag for states looking to curb prediction markets in their jurisdictions (see the last tech roundup for more on that fight):
A federal judge blocked Minnesota’s first-in-the-nation prediction-market ban days before it was set to take effect. In response, Gov. Tim Walz signed an executive order barring state employees from using insider information for prediction-market bets.
Days later, a federal judge ruled Utah can enforce its anti-gambling law against Kalshi, rejecting the company’s argument that federal commodities law preempts state authority.
Oregon Democrats proposed a three-year data center moratorium, joining 15 other states who are weighing similar pauses (New York is the only state to actually pass one).
Influence
Anthropic CEO Dario Amodei published a post denying that his company would like to ban open-weight models, an accusation that has been increasingly levied at the company. Instead, Amodei wrote Anthropic supports chip export controls and mandatory safety testing of all advanced models.
More than 1,300 employees at Anthropic, OpenAI, Meta, and Google signed an open letter calling on the U.S. government to help lead an international effort to build the technical and governance tools to deliberately slow the pace of frontier AI development.
That’s all for now. Have a great week!