Big Tech's preemption push: KOSA tied to AI bill before midterms
The White House floated packaging Sen. Marsha Blackburn’s Kids Online Safety Act with a federal AI preemption push.
TL;DR
- 01The White House floated packaging Sen. Marsha Blackburn’s Kids Online Safety Act with a federal AI preemption push.
- 02The administration reportedly told child safety groups and Big Tech that it would endorse Blackburn’s KOSA as part of an overall preemption package.
- 03For months lobbyists have pursued federal preemption, the idea of a single national AI law that would override state rules.
Big Tech lobbyists are pushing for a single federal AI preemption law, but a White House plan to pair that push with Senator Marsha Blackburn’s Kids Online Safety Act has created a scramble across Capitol Hill.
The administration reportedly told child safety groups and Big Tech that it would endorse Blackburn’s KOSA as part of an overall preemption package. That outreach leaked in early June 2026, and the move left House Republicans and some Democrats out of the loop, producing a week of confusion about which chamber’s version of KOSA would be used as the vehicle.
What happened
For months lobbyists have pursued federal preemption, the idea of a single national AI law that would override state rules. The White House released a draft of a comprehensive AI law in March of this year that already included some of the measures discussed in Washington. Leaked reports in early June indicated the White House had told groups it would back Blackburn’s Senate-backed KOSA as part of a preemption package.
The Senate version of KOSA, cosponsored by Senator Richard Blumenthal and overwhelmingly passed in 2024 by a 91–3 margin, would impose a “duty of care” on tech companies and extend responsibility to AI firms. The House has its own KOSA, led by Majority Leader Steve Scalise, and the House version diluted the duty-of-care provision late last November, prompting anger from child safety advocates.
That split left tech lobbyists and lawmakers asking which bill would win. As one Republican tech lobbyist said, “No one knows really who’s driving this thing.” The White House’s apparent preference for Blackburn’s bill further irritated House Republican leadership, which had passed a different KOSA, and left Democrats who helped craft the Senate bill surprised to learn it might be coupled to AI preemption.
Senate negotiations are complicated by numbers: a standalone bill moving through the Senate would face a 60-vote threshold. Observers noted that any fresh version of a KOSA-preemption package sent to the Senate would therefore require Democratic votes to clear cloture. The legislative calendar compounds the problem: it is mid June, there is roughly a month and a half before a five-week recess, and the runup to the general election season limits floor time. Lawmakers are already occupied with a slate of bills including the renewal of FISA, an immigration crackdown package, increased defense spending for the president’s war with Iran, a crypto market structure bill, affordability measures, the SAVE America election bill, and regular budget items like Medicaid.
The political coalitions and the Four Cs
The White House’s approach to preemption appears influenced by Mike Davis, a Trump-allied lawyer and founder of the Article III Project, who argued preemption must protect a set of values he calls the Four Cs: children, conservatives, creators, and communities. Davis told outlets he would block preemption that failed to address all four.
That framing helps explain why KOSA, which addresses children, surfaced as a potential companion to the preemption push. But it also underlines the trade-offs for Big Tech: accept a bundled package that includes tighter child-safety duties, or insist on a narrower preemption that avoids broad employer-style duties and state-level restrictions.
Why it matters
A bundled KOSA–preemption package would force tech companies to choose which concession matters more: a single federal AI regime or narrow language that weakens duty-of-care requirements. The timing makes trade-offs urgent; with the midterm calendar compressing legislative time, any complex compromise must clear deeply divided House and Senate coalitions in weeks rather than months. If Democrats win or consolidate power in either chamber after the election, the incentives to back a preemption deal negotiated with the current White House will collapse, according to lobbyists and advocates quoted by sources.
What to watch
Watch whether House Republican leaders sign on to Blackburn’s KOSA or insist on the Scalise version, and whether the White House formally transmits a combined KOSA–preemption bill to Congress. Also track whether a revised package can muster 60 votes in the Senate before the five-week recess and the general election calendar constrains floor time further.
- 2024Senate passage of KOSA
Senate KOSA, cosponsored by Senator Richard Blumenthal, passed 91–3 in 2024.
- March 2026White House releases AI draft
The White House released a draft of a comprehensive AI law in March of this year that included some protections aligned with the Four Cs.
- Early June 2026Leaked outreach on KOSA
Reports leaked that the White House told child safety groups and Big Tech it would endorse Blackburn’s KOSA as part of an AI preemption package.
- Mid June 2026Capitol Hill confusion and timing crunch
House Republicans and some Democrats said they had not been informed; observers noted roughly a month and a half before a five-week recess and a crowded legislative calendar.
Written by The Brieftide · Source: The Verge
The Brieftide Daily · 06:00
Briefs like this one, in your inbox every morning.
Continue reading
More in AI SafetyAnthropic's Dario Amodei publishes 'Policy on the AI Exponential'
Amodei paired a long essay with two frameworks calling for mandatory third-party audits, agency power to block models.
German AI Security Institute greenlit and modeled on UK AISI
Germany's National Security Council approved the institute to test advanced AI risks.
Google DeepMind funds up to $10M for multi-agent safety
DeepMind and partners are offering a global research call on multi-agent AI safety, with a deadline of August 8, 2026.
OpenAI shifts from full automation: human-AI 'tandem' by 2028
Sam Altman and Jakub Pachocki say by March 2028 a significant fraction of OpenAI's research may be done in human-AI tandems and propose an.