The Kids Online Safety Act is back. Again.
KOSA — a bill that has failed to pass multiple times because civil liberties organizations keep pointing out it’s a censorship bill wrapped in a child safety bow — has new friends this time around. Big ones. And the price of their support should make every internet user furious.
In May, OpenAI publicly endorsed KOSA. Not because Sam Altman suddenly cares about protecting teenagers. Because OpenAI’s lobbyists are cutting a deal: support KOSA in exchange for Congress going easy on AI regulation. Trade your free speech for their market dominance. Everyone wins — except you.
What KOSA Actually Does
KOSA — the Kids Online Safety Act — would require online platforms to “exercise reasonable care” to prevent a long list of harms to minors. The list includes depression, anxiety, eating disorders, substance use, bullying, and something called “compulsive usage.”
That sounds reasonable until you think about it for thirty seconds.
“Compulsive usage” is not a medical diagnosis. There is no scientific consensus that online platforms cause the mental health disorders KOSA lists. The term was invented for this legislation — an undefined concept given legal teeth, designed to sound like settled science.
Under KOSA, if a teenager reads a forum post about managing depression and later has a mental health crisis, the platform that hosted that forum can be sued by the FTC or any state attorney general in the country. The platform doesn’t have to have done anything wrong. It just has to have hosted content that someone later decides was “harmful.”
What happens next is predictable: platforms delete the forums. They remove the support groups. They block content about eating disorders, drug abuse, self-harm, and mental health — not because the content is dangerous, but because talking about these topics at all becomes a legal liability.
KOSA doesn’t protect kids from harm. It protects platforms from lawsuits — by making them censor the exact communities where kids go for help.
The LGBTQ+ Target on KOSA’s Back
This isn’t hypothetical. KOSA’s co-sponsor, Senator Marsha Blackburn of Tennessee, has been explicit about what she considers “harmful content” for children: LGBTQ+ identity.
Blackburn has publicly stated she wants KOSA used to silence trans voices online. Her co-sponsor, Senator Richard Blumenthal of Connecticut, claims the bill has no First Amendment problems — and then, in January 2025, admitted on the record that KOSA is designed to “stop destructive content on bullying, eating disorders, and self-harm.”
Stop content. His word, not ours.
So one co-sponsor wants to use the bill to censor trans people, and the other admits it’s designed to stop content he considers destructive. And they’re both claiming it doesn’t threaten free speech.
We wrote last month about how the White House’s 2026 Counterterrorism Strategy labeled pro-transgender advocacy as a national security threat. KOSA would hand the enforcement mechanism to every state attorney general in the country. Imagine a Texas AG deciding that a support forum for trans teenagers constitutes “harmful content” and suing the platform into silence. That’s not a slippery slope argument — it’s the stated design of the bill.
The AI Deal: Your Free Speech for Their Regulatory Moat
Now here’s where it gets ugly.
OpenAI endorsed KOSA in May 2026 through Chris Lehane, their Chief Global Affairs Officer. Lehane is not a technologist. He’s not a child safety expert. He’s a political operator — a professional dealmaker who spent decades in Democratic Party politics before joining OpenAI.
You don’t roll out Chris Lehane to take a principled position on child safety. You roll him out to cut a deal.
The deal is simple: KOSA primarily targets social media platforms, not AI companies. The compliance costs — content moderation, age verification, legal liability for user-generated content — will be ruinous for small platforms and startups, but manageable for companies like OpenAI that don’t host user-generated social feeds. OpenAI endorses KOSA, Congress gets bipartisan cover for a “kids safety” bill, and in return, the AI companies get favorable treatment when AI-specific regulation comes up.
As Techdirt put it: “It’s regulatory capture with a smiley face.”
OpenAI isn’t the only one. Apple and X (formerly Twitter) have also endorsed KOSA. X actually helped negotiate the text of a previous version. These are not companies acting out of concern for children. These are companies that know KOSA will hurt their smaller competitors more than it hurts them.
Meanwhile, Blumenthal used the OpenAI endorsement to claim that the only opposition left to KOSA is “Mark Zuckerberg & his lobbyists.” That’s a lie. Here’s who actually opposes KOSA:
- The Electronic Frontier Foundation
- The ACLU
- Over 90 civil society and public interest organizations
- Senator Rand Paul
- Digital rights groups across the political spectrum
But “90+ civil liberties organizations oppose our bill” doesn’t make as good a tweet as “only Zuckerberg is against us.”
We’ve Seen This Movie Before: FOSTA/SESTA
KOSA isn’t the first bill sold to Congress under “think of the children.” In 2018, Congress passed FOSTA/SESTA — the Allow States and Victims to Fight Online Sex Trafficking Act — with the promise it would stop child sex trafficking online.
It didn’t work. Independent studies found FOSTA/SESTA was ineffective at reducing sex trafficking. What it actually did was push sex workers off safer online platforms and onto the street, eliminate harm-reduction communities, and make trafficking victims harder to find. Even the politicians who originally championed FOSTA/SESTA now acknowledge it failed.
KOSA follows the same playbook: exploit genuine parental fear, write a vague law with broad enforcement powers, hand those powers to politically motivated officials, and watch as the collateral damage hits the most vulnerable communities first. The EFF has been explicit that KOSA will likely make kids less safe by destroying the online communities where they find support.
What the Pirate Party Says
The California Pirate Party analyzes policy through what we call the Pirate Wheel — eight principles that any law must respect to be legitimate. KOSA fails nearly all of them:
- Privacy — our principle that individuals own their data and identity. KOSA will inevitably require age verification to function, meaning platforms must collect and verify users’ real identities. That’s a national surveillance infrastructure built to “protect children.”
- TICKS (Tools, Ideas, Culture, Knowledge, Sentiments) — our principle that any communicable information should flow freely. KOSA directly restricts legal speech about health, identity, and harm reduction — the exact kind of information that saves lives.
- Humanism — our principle of equal rights regardless of born traits. KOSA disproportionately targets LGBTQ+ communities, and its co-sponsor has said so explicitly. A law that enables selective censorship of minority voices fails this principle completely.
- Quality Legislation — our requirement that laws be necessary, effective, proportionate, and evidence-based. KOSA fails all four: there’s no scientific consensus that platforms cause the listed harms, no evidence KOSA would reduce them, and the collateral damage to free speech is wildly disproportionate.
- Resilience — our principle against single points of failure and drift toward authoritarianism. KOSA centralizes content control in the hands of government enforcers and the largest corporations. Small platforms die. Independent communities vanish. The internet becomes a walled garden run by the same companies lobbying for the bill.
- The core asymmetry — our foundational principle that privacy belongs to individuals and transparency belongs to institutions. KOSA inverts this: it demands individuals prove their age (destroying privacy) while allowing corporations and lawmakers to cut deals behind closed doors about whose speech gets protected and whose gets censored.
What You Can Do
KOSA has not yet received a floor vote in the 119th Congress, but its sponsors are building momentum. Here’s how to push back:
- Contact your senators — Tell them you oppose KOSA. The EFF has a direct action page at act.eff.org. It takes two minutes. California’s senators need to hear from constituents, not just lobbyists.
- Don’t fall for the framing — When someone says “don’t you care about kids?”, the answer is: “Yes, and that’s why I oppose a bill that will destroy the online support communities kids depend on.” KOSA’s sponsors count on the emotional weight of child safety to shut down criticism. Don’t let them.
- Follow the money — When a company like OpenAI endorses a “child safety” bill that primarily hurts their competitors, that’s not altruism. That’s regulatory capture. Name it.
- Read the EFF’s analysis — Their in-depth breakdown of KOSA’s constitutional problems is the best single resource. Share it.
- Remember FOSTA/SESTA — The last time Congress passed a “think of the children” internet bill, it made everything worse. Ask your representatives: what’s different this time? (The answer is: nothing.)
Censoring the internet doesn’t protect children. It never has. What protects children is comprehensive privacy legislation that stops platforms from harvesting their data, strong community moderation run by actual humans, and parents who are engaged in their kids’ lives — not a vague federal “duty of care” enforced by state attorneys general with political agendas.
KOSA is not a child safety bill. It’s a censorship bill endorsed by trillion-dollar companies because it eliminates their competition. Congress should reject it, and every organization that endorsed it should be asked one question: who are you actually protecting?
The California Pirate Party stands for a free and open internet — for everyone, including kids. We oppose KOSA because we believe children deserve access to information, community, and support, not a sanitized internet curated by politicians and corporations.
We keep us safe.
The Kids Online Safety Act is back. Again.
KOSA — a bill that has failed to pass multiple times because civil liberties organizations keep pointing out it’s a censorship bill wrapped in a child safety bow — has new friends this time around. Big ones. And the price of their support should make every internet user furious.
In May, OpenAI publicly endorsed KOSA. Not because Sam Altman suddenly cares about protecting teenagers. Because OpenAI’s lobbyists are cutting a deal: support KOSA in exchange for Congress going easy on AI regulation. Trade your free speech for their market dominance. Everyone wins — except you.
What KOSA Actually Does
KOSA — the Kids Online Safety Act — would require online platforms to “exercise reasonable care” to prevent a long list of harms to minors. The list includes depression, anxiety, eating disorders, substance use, bullying, and something called “compulsive usage.”
That sounds reasonable until you think about it for thirty seconds.
“Compulsive usage” is not a medical diagnosis. There is no scientific consensus that online platforms cause the mental health disorders KOSA lists. The term was invented for this legislation — an undefined concept given legal teeth, designed to sound like settled science.
Under KOSA, if a teenager reads a forum post about managing depression and later has a mental health crisis, the platform that hosted that forum can be sued by the FTC or any state attorney general in the country. The platform doesn’t have to have done anything wrong. It just has to have hosted content that someone later decides was “harmful.”
What happens next is predictable: platforms delete the forums. They remove the support groups. They block content about eating disorders, drug abuse, self-harm, and mental health — not because the content is dangerous, but because talking about these topics at all becomes a legal liability.
KOSA doesn’t protect kids from harm. It protects platforms from lawsuits — by making them censor the exact communities where kids go for help.
The LGBTQ+ Target on KOSA’s Back
This isn’t hypothetical. KOSA’s co-sponsor, Senator Marsha Blackburn of Tennessee, has been explicit about what she considers “harmful content” for children: LGBTQ+ identity.
Blackburn has publicly stated she wants KOSA used to silence trans voices online. Her co-sponsor, Senator Richard Blumenthal of Connecticut, claims the bill has no First Amendment problems — and then, in January 2025, admitted on the record that KOSA is designed to “stop destructive content on bullying, eating disorders, and self-harm.”
Stop content. His word, not ours.
So one co-sponsor wants to use the bill to censor trans people, and the other admits it’s designed to stop content he considers destructive. And they’re both claiming it doesn’t threaten free speech.
We wrote last month about how the White House’s 2026 Counterterrorism Strategy labeled pro-transgender advocacy as a national security threat. KOSA would hand the enforcement mechanism to every state attorney general in the country. Imagine a Texas AG deciding that a support forum for trans teenagers constitutes “harmful content” and suing the platform into silence. That’s not a slippery slope argument — it’s the stated design of the bill.
The AI Deal: Your Free Speech for Their Regulatory Moat
Now here’s where it gets ugly.
OpenAI endorsed KOSA in May 2026 through Chris Lehane, their Chief Global Affairs Officer. Lehane is not a technologist. He’s not a child safety expert. He’s a political operator — a professional dealmaker who spent decades in Democratic Party politics before joining OpenAI.
You don’t roll out Chris Lehane to take a principled position on child safety. You roll him out to cut a deal.
The deal is simple: KOSA primarily targets social media platforms, not AI companies. The compliance costs — content moderation, age verification, legal liability for user-generated content — will be ruinous for small platforms and startups, but manageable for companies like OpenAI that don’t host user-generated social feeds. OpenAI endorses KOSA, Congress gets bipartisan cover for a “kids safety” bill, and in return, the AI companies get favorable treatment when AI-specific regulation comes up.
As Techdirt put it: “It’s regulatory capture with a smiley face.”
OpenAI isn’t the only one. Apple and X (formerly Twitter) have also endorsed KOSA. X actually helped negotiate the text of a previous version. These are not companies acting out of concern for children. These are companies that know KOSA will hurt their smaller competitors more than it hurts them.
Meanwhile, Blumenthal used the OpenAI endorsement to claim that the only opposition left to KOSA is “Mark Zuckerberg & his lobbyists.” That’s a lie. Here’s who actually opposes KOSA:
But “90+ civil liberties organizations oppose our bill” doesn’t make as good a tweet as “only Zuckerberg is against us.”
We’ve Seen This Movie Before: FOSTA/SESTA
KOSA isn’t the first bill sold to Congress under “think of the children.” In 2018, Congress passed FOSTA/SESTA — the Allow States and Victims to Fight Online Sex Trafficking Act — with the promise it would stop child sex trafficking online.
It didn’t work. Independent studies found FOSTA/SESTA was ineffective at reducing sex trafficking. What it actually did was push sex workers off safer online platforms and onto the street, eliminate harm-reduction communities, and make trafficking victims harder to find. Even the politicians who originally championed FOSTA/SESTA now acknowledge it failed.
KOSA follows the same playbook: exploit genuine parental fear, write a vague law with broad enforcement powers, hand those powers to politically motivated officials, and watch as the collateral damage hits the most vulnerable communities first. The EFF has been explicit that KOSA will likely make kids less safe by destroying the online communities where they find support.
What the Pirate Party Says
The California Pirate Party analyzes policy through what we call the Pirate Wheel — eight principles that any law must respect to be legitimate. KOSA fails nearly all of them:
What You Can Do
KOSA has not yet received a floor vote in the 119th Congress, but its sponsors are building momentum. Here’s how to push back:
Censoring the internet doesn’t protect children. It never has. What protects children is comprehensive privacy legislation that stops platforms from harvesting their data, strong community moderation run by actual humans, and parents who are engaged in their kids’ lives — not a vague federal “duty of care” enforced by state attorneys general with political agendas.
KOSA is not a child safety bill. It’s a censorship bill endorsed by trillion-dollar companies because it eliminates their competition. Congress should reject it, and every organization that endorsed it should be asked one question: who are you actually protecting?
The California Pirate Party stands for a free and open internet — for everyone, including kids. We oppose KOSA because we believe children deserve access to information, community, and support, not a sanitized internet curated by politicians and corporations.
We keep us safe.