THE LAST AMENDMENT
HOW THE PEOPLE WHO OWN THE INTERNET ARE TRYING TO OWN THE RULES OF IT, AND WHAT HAPPENS TO YOU WHEN THEY SUCCEED
“When you trade freedom for security, you deserve neither and will lose both.” — Benjamin Franklin (widely attributed, variants in his writings)
“If liberty means anything at all, it means the right to tell people what they do not want to hear.” — George Orwell, proposed preface to Animal Farm, 1945
“The internet interprets censorship as damage and routes around it.” — John Gilmore, co-founder of the Electronic Frontier Foundation, 1993
“Whoever would overthrow the liberty of a nation must begin by subduing the freeness of speech.” — Benjamin Franklin, writing as “Silence Dogood,” 1722
“The first thing a government does when it takes control is take control of the press.” — Attributed to Napoleon Bonaparte
“There is no such thing as a little freedom. Either you are all free, or you are not free.” — Walter Cronkite
WHAT SECTION 230 IS, WHERE IT CAME FROM, AND WHY TWENTY-SIX WORDS ARE THE ONLY THING STANDING BETWEEN YOU AND A VERY DIFFERENT INTERNET
In the summer of 1995, a judge handed down a ruling in a case called Stratton Oakmont v. Prodigy Services that should have killed the internet before most Americans had ever seen it. Prodigy, an early online service, had made the mistake of moderating some of the content on its message boards. A court decided that by moderating anything, Prodigy had made itself legally responsible for everything — every post, every comment, every word typed by every user on its platform. The ruling made the logic explicit: if you curate, you are a publisher, and publishers are liable for what they publish.
The implication was simple and catastrophic. Any online platform that attempted to remove harmful content — child exploitation material, defamation, harassment — thereby accepted legal liability for all content it failed to remove. The only rational business response was to moderate nothing, ever, and let the sewers flow unchecked. The alternative was bankruptcy via lawsuit.
Two members of Congress saw the ruling and understood immediately what it would do. Representative Chris Cox of California, a Republican, and Representative Ron Wyden of Oregon, a Democrat, wrote a twenty-six word amendment to a bill already moving through Congress. Those twenty-six words are:
“No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”
That is Section 230 of the Communications Decency Act of 1996. Cox later described its intent as twofold: to protect speech and privacy on the internet from government regulation, and to create legal space for platforms to moderate harmful content without becoming liable for all content. The amendment passed as part of the Telecommunications Act of 1996, signed by President Clinton on February 8, 1996.
It is not hyperbole to say that Section 230 built the internet as it currently exists. Without it, Reddit does not exist. Wikipedia does not exist. Yelp does not exist. YouTube does not exist. The comment section on every news website, the ability to post on any platform, the entire architecture of user-generated content online — all of it rests on the legal foundation that the platform is not responsible for what its users say.
Section 230 has two functional components. The first shields platforms from liability for third-party content. The second explicitly protects their right to moderate — to remove content in good faith — without that moderation being used as evidence that they are now publishers responsible for everything else. It is, simultaneously, a protection for speech and a protection for the curation of speech. Cox and Wyden deliberately designed it that way, because they understood that a platform forced to choose between moderation and survival would choose survival, and that choice would produce something worse than either option.
Section 230 does not protect illegal content. It does not shield platforms that create illegal content or are directly involved in it. It protects passive hosting of content created by others. It is the legal equivalent of the post office not being responsible for the contents of the letters it delivers.
For thirty years, despite pressure from both parties for different and often contradictory reasons, it has held.
HOW THE INTERNET WORKS RIGHT NOW
The internet in 2026 is not a free speech utopia. Anyone who has had a tweet removed, a Facebook post fact-checked, a YouTube video demonetized, or a Reddit account banned knows that platforms exercise enormous power over what is and is not visible. That power is exercised through algorithms that are proprietary, opaque, and demonstrably shaped by both commercial incentives and, increasingly, political pressure.
Engagement-maximizing algorithms amplify outrage because outrage produces clicks and clicks produce advertising revenue. Content moderation policies are written by teams whose decisions are not subject to judicial review, administrative process, or public comment. A platform can remove your account, your livelihood, and your ability to communicate with the communities you have built, for reasons that are not fully explained and cannot be effectively appealed.
This is the internet’s existing compromise with free speech. It is not good. It is deeply unequal. The powerful have more recourse than the powerless. The rich can hire lawyers. The verified accounts get human review. The anonymous, the marginal, and the inconvenient get the algorithm.
And yet — and this is the critical point — the existing system is not totalitarian. A hundred different platforms operate under different policies. Content removed from one appears on another. Substack exists because Medium made choices its writers disagreed with. Mastodon exists because Twitter made choices its users disagreed with. The proliferation of alternatives, however imperfect, is a structural feature of a decentralized system in which no single actor controls all channels of communication simultaneously.
Section 230 is the legal foundation of that decentralization. It is what makes it economically possible for small platforms to exist at all, because without its liability shield, only entities with the legal resources to defend against any conceivable lawsuit could afford to operate. Without Section 230, the internet consolidates into the three or four platforms large enough to absorb litigation risk. And those three or four platforms — already large enough to be political targets, already susceptible to government pressure — become the only game in town.
The current system is a flawed compromise. The proposed replacement is a controlled bottleneck.
WHAT COMES NEXT, AND WHY YOU SHOULD BE ALARMED
On December 11, 2025, President Trump signed an executive order calling for a unified federal approach to AI regulation that would preempt — that is, override and nullify — state-level AI laws across the country. On March 18, 2026, Senator Marsha Blackburn of Tennessee released a 300-page discussion draft of legislation she titled, without apparent irony, the “Republic Unifying Meritocratic Performance Advancing Machine Intelligence by Eliminating Regulatory Interstate Chaos Across American Industry Act.” The acronym spells TRUMP AMERICA AI Act. This is a bill named after a person. The bill would do the following things.
First, it would sunset — eliminate — Section 230 two years after passage. Every platform in the United States would become legally responsible for every piece of user-generated content it hosts, or face unlimited civil liability. The practical result is one of two outcomes: platforms stop allowing user-generated content altogether, or platforms implement automated surveillance of all content at the point of creation, removing anything that might conceivably trigger legal liability before it is ever published.
Second, it would preempt state AI regulations — meaning that the 1,208 AI-related bills introduced across state legislatures in 2025, of which 145 were enacted into law, would be nullified. States would lose the authority to regulate AI systems operating within their borders in areas covered by the federal framework. Colorado’s AI Act, which required bias assessments and consumer disclosures for high-risk AI systems making consequential decisions about employment, healthcare, and housing, would be overridden. California’s privacy-protective AI regulations would be overridden.
The White House is currently negotiating with Congress to bundle the AI preemption provision with three popular child safety bills — the Kids Online Safety Act, the No Fakes Act, and a federal age verification mandate — in order to provide enough bipartisan cover to pass a package that Congress has already rejected twice. The Senate voted 99-1 to strip AI preemption language from the One Big Beautiful Bill Act earlier this year. Seventeen Republican governors joined the opposition. The administration, having failed twice through the legislative process, is now attempting to pass preemption by attaching it to legislation that is difficult to vote against publicly.
To understand what this means in practice, consider the sequence: a child safety bill is proposed. The preemption clause is bundled with it. Any senator who votes against the package is now on record as voting against child safety. The preemption clause passes. Every state AI regulation in the country is nullified. The federal standard — written in consultation with the White House, with input from the tech industry, subject to none of the accountability mechanisms that produced those state laws — becomes the only law.
The administration has also deployed an additional weapon: it has threatened to condition $42 billion in broadband infrastructure funding — money already allocated by Congress to close the digital divide in rural and underserved communities — on states agreeing not to enforce AI regulations the White House considers “burdensome.” This is extortion with a federal grant attached. States that refuse to abandon their AI accountability laws lose the broadband money. The people who lose that connectivity are not the tech executives who lobbied for preemption. They are the people in rural Tennessee and rural Washington who still rely on satellite internet and whose senators are voting to protect the industry that provides it.
THE DESTINATION, STATED PLAINLY
The endpoint of removing Section 230, eliminating state AI oversight authority, and replacing both with a federal standard written to the specifications of the tech industry is a surveillance-ready internet in which the platforms are both legally incentivized to monitor all content and legally insulated from state-level accountability for how they do it.
Under this architecture, every post, every message, every search query, every image, every video is subject to automated review by systems built and operated by companies whose data infrastructure is already integrated with federal surveillance systems via contracts worth hundreds of millions of dollars annually. The company that holds the majority of those contracts is Palantir Technologies. Palantir’s federal contracts have grown from $4.4 million in 2009 to $970 million in 2024. Since Trump’s inauguration in January 2025, Palantir has been awarded more than $1.3 billion in federal contracts across the Departments of Defense, Homeland Security, Treasury, State, Health and Human Services, Veterans Affairs, Energy, Transportation, Justice, Agriculture, HUD, and Commerce.
Palantir does not collect the data. It is the system through which collected data becomes searchable, cross-referenceable, and actionable at scale. Its ICE contract, worth $30 million, provides near-real-time location visibility for deportation targeting. The Trump administration has used Palantir to build a master database cross-referencing tax records, immigration records, and Social Security data. The company is, as Wired described it, becoming “an operating system for the entire government.”
The man who co-founded Palantir, named it after a surveillance device from a fantasy novel, seeded it with money from the CIA’s venture capital arm, and now chairs its board is Peter Thiel. Thiel has also, over the course of two decades, funded the political careers of multiple senators and representatives, personally lobbied the President of the United States to install his protégé as Vice President, co-founded the Dialog Society — the invitation-only network whose 2026 agenda includes sessions titled “Build-a-Cult” and “Navigating WWIII” and whose attendees include the sitting Treasury Secretary and the Supreme Allied Commander of NATO — and maintained a documented, years-long correspondence with Jeffrey Epstein, including discussions about investment arrangements and what Epstein described in a February 2016 email as Thiel’s interest in a “secret society idea.”
The Nation described the Epstein-Thiel relationship as a “deep, symbiotic friendship” documented across hundreds of email exchanges in the released federal files. Epstein invested $40 million into Valar Ventures, a firm Thiel co-founded. Their documented correspondence began in 2014, six years after Epstein’s Florida sex-crimes conviction, and continued until at least 2019. Thiel was present at the August 2015 Palo Alto dinner — documented in the federal Epstein files via Epstein’s own email to himself the following morning — that also included Elon Musk and Mark Zuckerberg.
Neither Thiel nor Musk has been accused of criminal conduct in connection with Epstein. That is a factual statement. It is also a factual statement that both men continued to maintain documented relationships with a convicted sex offender for years after his conviction, and that both have declined to fully account for those relationships publicly.
THE PLAYERS: WHO IS PUSHING THIS, WHO IS OPPOSING IT, AND WHO IS PAYING FOR WHAT
FOR federal preemption and Section 230 repeal:
Senator Marsha Blackburn (R-TN) is the primary legislative sponsor of the TRUMP AMERICA AI Act. She is also currently running for governor of Tennessee. Her bill was written in direct coordination with the White House, per Axios reporting on the drafts. Backers of her proposal include the RIAA (Recording Industry Association of America), MPA (Motion Picture Association), and the Institute for Family Studies — the collection of industry and conservative organizations whose primary interest in AI regulation is copyright protection and content control, not consumer protection.
The White House AI framework was released two days after Blackburn’s draft, on March 20, 2026. It endorses preemption but is notably silent on Section 230 repeal, suggesting internal disagreement within the administration about how far to go. The executive order signed in December 2025 cited Colorado’s AI accountability law as an example of a regulation that “may even force AI models to produce false results” — a characterization that Colorado lawmakers disputed directly and that consumer groups called misleading.
The Foundation for American Innovation, a tech-industry-aligned think tank, called the Blackburn bill “a disaster” for technological leadership, which represents the Silicon Valley wing of the Republican coalition breaking from the social conservative wing over how much regulation is tolerable when it cuts both ways.
Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) have introduced a separate bill, the Great American Artificial Intelligence Act, which would impose a federal ban on state AI regulation. It is the third preemption attempt in as many years.
AGAINST federal preemption:
The ACLU has issued a formal statement calling the Blackburn bill’s preemption provisions a failure that “block states from enforcing existing state protections and from enacting new protections.” This could include anything from privacy regulations to anti-discrimination requirements to AI safety laws.
More than 130 civil society organizations — including SEIU, the Center for Biological Diversity, Common Cause, Consumer Federation of America, Public Citizen, and UltraViolet Action — signed a joint letter organized by Demand Progress asking congressional leaders of both parties to reject the preemption legislation.
Seventeen Republican governors joined bipartisan opposition to the first preemption attempt. The Senate voted 99-1 against it. The administration came back with a different packaging. It will keep coming back until it wins or loses a president.
Ron Wyden, the Democratic senator from Oregon and the co-author of Section 230, has opposed its repeal consistently. He has argued that repealing Section 230 without replacing it with something coherent would hand enormous new power to the largest, most legally fortified platforms while destroying smaller competitors who cannot afford the litigation exposure.
States have not waited for federal permission. As of June 2026, 1,208 AI bills have been introduced in state legislatures, and 145 have been enacted. Colorado passed and then narrowed its AI governance law. California has enacted multiple AI transparency and safety provisions. Illinois has AI employment rules. These are the laws that preemption would nullify.
THE MONEY:
Peter Thiel has invested approximately $1.5 million in pro-Trump political infrastructure in 2016, $15 million to J.D. Vance’s 2022 Ohio Senate campaign (described by Politico as “the largest amount ever given to boost a single Senate candidate”), and $1.7 million in federal political contributions during the 2024 cycle. He has done so while his company Palantir receives federal contracts worth more than $1.3 billion since Trump’s second inauguration, with that figure growing.
The Supreme Court’s 2010 Citizens United decision made the financial architecture supporting all of this legally possible. By declaring corporate political spending a form of protected speech and eliminating meaningful limits on outside political expenditure, Citizens United created the legal environment in which a billionaire can simultaneously fund a candidate’s political career, lobby that candidate’s administration for policy outcomes, and receive hundreds of millions in government contracts from the administration that candidate runs — without any of these transactions being illegal.
Dark money from groups that do not disclose their donors — spending that was largely illegal before Citizens United — topped nearly $2 billion in the 2024 election cycle, roughly double the total from 2020. The Brennan Center’s analysis found that the real total is almost certainly higher. The donors behind that money are, by design, unknown.
Senator Dick Durbin noted in January 2026 that crypto industry donors spent $40 million to defeat a single Democratic senator who had taken positions against the industry. That is not a lobbying effort. That is a targeted electoral execution, funded anonymously, made legal by a 2010 Supreme Court decision in which the majority opinion stated, without evidence, that unlimited corporate political spending would not “give rise to corruption or the appearance of corruption.”
Citizens United did not create the oligarchic capture of American democratic institutions. It accelerated and legalized it. The capture began earlier, with the systematic deregulation of campaign finance that followed the Reagan administration’s abandonment of the Fairness Doctrine in 1987 — the FCC rule that required broadcasters to present contrasting views on public issues, and whose repeal, within five years, produced Rush Limbaugh, Fox News, and the media ecosystem that made the current political moment possible. The Fairness Doctrine was not killed by Congress. It was killed by an FCC chairman acting under executive pressure. The FCC chairman today is Brendan Carr. The pattern is not new.
THE EPSTEIN THREAD Leads TO…
The question of whether there is an Epstein legacy in the current legislative push for AI preemption and Section 230 repeal is not a question about criminal conspiracy. No one is arguing that Marsha Blackburn’s 300-page bill was dictated in a private island mansion. The question is a simpler one: whether the network of financial relationships, personal connections, and shared institutional interests that Epstein cultivated for three decades has continued to function after his ‘death’ as a connective tissue between money, power, and policy — and whether the outcomes that network produces are structurally consistent with what Epstein was doing while he was alive.
Consider what Epstein did. He collected relationships with the powerful. He made himself financially useful to people who had more money than they needed in ways that are not fully documented. He maintained documented correspondence with Peter Thiel beginning in 2014 and continuing to 2019, including discussions of investment arrangements, shared interest in AI and life extension, and what Thiel described to the Rogan podcast audience as a friendship he has not fully explained. He invested $40 million into a firm Thiel co-founded. He discussed with Thiel, in a documented email, the concept of a “secret society.”
Peter Thiel co-founded Dialog in 2006 — the same year Epstein was first arrested in Florida. Epstein’s name appeared on Dialog’s 2014 invitation list. Thiel has built, over twenty years, the most consequential privately-held surveillance infrastructure in the United States, which is now integrated into the operational systems of twelve or more federal departments. He has also installed, via a $15 million investment and a personal introduction to the former president, a Vice President of the United States who once privately described Trump as “an American Hitler” before deciding that association with Trump’s political infrastructure was consistent with his? ambitions.
The people asking Congress to eliminate state AI oversight authority and to repeal Section 230 are not doing so because they want less power over the digital infrastructure of American life. They are doing so because the existing patchwork of state laws — 1,208 bills in one year, 145 enacted — represents the last active layer of democratic accountability over systems that will, within a decade, make decisions about your employment, your healthcare, your insurance, your housing, your immigration status, and your access to credit. Eliminating that layer does not create a power vacuum. It transfers the power that layer currently exercises to the entities that will operate in the absence of it.
Those entities are the companies that hold the federal contracts, maintain the data infrastructure, and fund the political campaigns of the people writing the legislation that governs them.
HOW WE GOT HERE, AND WHY IT MATTERS THAT WE UNDERSTAND THE SEQUENCE
1787: The Constitution is ratified with no Bill of Rights. The Bill of Rights is added in 1791 because the founders understood, from direct experience, that a government without explicit constraints on its authority over speech, assembly, and due process would expand those constraints at the expense of the people it governed.
1798: The Alien and Sedition Acts make it a crime to publish “false, scandalous, and malicious writing” against the government. They are used to prosecute newspaper editors who criticize President Adams. They expire in 1801 and are not renewed because they are catastrophically unpopular. The lesson — that government will use any legal tool available to suppress its critics — is noted and largely ignored in subsequent generations.
1919: Justice Oliver Wendell Holmes writes the “clear and present danger” standard in Schenck v. United States, upholding the conviction of a man who distributed anti-draft pamphlets during World War I. The same year, Holmes dissents in Abrams v. United States, writing that “the best test of truth is the power of the thought to get itself accepted in the competition of the market.” The First Amendment jurisprudence that protects American speech today descends more from that dissent than from the majority opinion.
1987: The FCC eliminates the Fairness Doctrine under Reagan. Within five years, commercial talk radio transforms American political discourse. Within a decade, Fox News launches. Within fifteen years, the media ecosystem that made the election of Donald Trump possible is fully operational.
2010: Citizens United v. FEC. Corporate political spending is declared protected speech. Super PACs are created. Dark money becomes institutionalized. The fusion of private wealth and electoral power that the founders explicitly feared becomes the operating system of American democracy.
2016: Peter Thiel donates $1.5 million to Trump and speaks at the Republican National Convention. J.D. Vance is working at Mithril Capital, Thiel’s venture fund. Epstein is alive, corresponding regularly with Thiel, and not in jail.
2019: Epstein ‘dies’ in federal custody. The surveillance cameras outside his cell are not functioning. The guards are asleep. The medical examiner rules it suicide. A forensic pathologist hired by his family rules it homicide. No one is charged.
2022: Thiel gives $15 million to J.D. Vance’s Senate campaign, the largest single-candidate donation of his career. Vance wins.
2024: Thiel introduces Vance to Trump. Trump names Vance as his running mate. The ticket wins.
2025: Palantir receives more than $1.3 billion in federal contracts across twelve or more agencies. The Trump administration uses Palantir to build a master database of federal records. The FCC, under Brendan Carr, begins threatening broadcast licenses and targeting late-night comedians. The DOGE operation, partly running in parallel with Palantir’s federal data consolidation, systematically removes the civil servants most likely to resist it.
2026: Marsha Blackburn introduces a 300-page bill to eliminate state AI regulation and repeal Section 230. The White House negotiates to bundle preemption with child safety legislation. The Senate has already rejected preemption 99-1. The administration tries again.
THE RATS, STRIPPED DOWN
Marsha Blackburn is the face on the legislation. She is running for governor of Tennessee and is using the TRUMP AMERICA AI Act to establish her credentials as a Trump loyalist before the primary. The bill is named after the President of the United States. This is not a coincidence. It is a political calculation. She is not the architect of the policy. She is the vehicle for it.
Peter Thiel is the architect. He has spent twenty years building the surveillance infrastructure, funding the political careers, organizing the private networks, and lobbying the regulatory agencies that together constitute a coherent long-term project: the transfer of digital power from decentralized, state-accountable systems to centralized, federally-contracted ones in which the primary beneficiary is the company he co-founded and chairs. He has done this while describing himself as a libertarian. He has also exchanged hundreds of emails with a convicted sex trafficker and alleged holder of kompromat on the rich and powerful “Epstein Class” for five years after that trafficker’s conviction, invested that trafficker’s money, discussed “secret society” concepts with him, and declined to fully explain any of it. The Nation describes their relationship as a “deep, symbiotic friendship.” Epstein described Thiel as a “great friend.” Epstein’s files show that Thiel’s name was on the Dialog invitation list — the same Dialog whose 2026 agenda includes “Build-a-Cult” and “Navigating WWIII.”
J.D. Vance is the most powerful protégé Thiel has produced. He is also one heartbeat from the presidency. He has no independent political base, no legislative record of consequence, and no explanation for how a man who was unknown outside Appalachian memoir circles in 2015 became Vice President of the United States in 2025 that does not involve Peter Thiel’s $15 million and Peter Thiel’s personal relationship with Donald Trump. He is a brand, deployed to provide a working-class face on a policy project that benefits the class of people who funded his career- the “Epstein Class”.
Elon Musk is the infrastructure. Starlink provides the satellite internet on which rural America increasingly depends. SpaceX has contracts with the federal government that dwarf almost any other private space company. Tesla’s self-driving (sort of) systems are one of the largest active AI deployments in consumer transportation. X (formerly Twitter) is the primary real-time public communications platform for American political discourse, owned by a man who was present at the 2015 Epstein dinner, who has not explained that presence, and who communicated with Epstein via Signal — a channel that produces no records — per CBS (not suss at all) News reporting on gaps in the released files.
Les Wexner is the cautionary example of what happens when people with unlimited money decide that the rules do not apply to them, find someone willing to operate in that space, and then, when the body of evidence against them reaches congressional scrutiny, say they were naive. He is 88 years old and his name appears more than 1,000 times in federal Epstein files. His name is on a congressional deposition that produced, from a Republican congressman, the word “liar.” His attorney was caught on a hot mic saying “I will fucking kill you.” He gave a man he describes as a professional acquaintance unlimited power of attorney over his fortune for fifteen years. He is not in jail. He is in Ohio. He says he was foolish. I guess he likes to jerk people off : ( !!!
Citizens United is not, as the question implies, the original cause. It is a tool. The original cause is the organized and deliberate capture of the regulatory, legislative, and judicial systems by capital over a period of decades, beginning with the Powell Memorandum of 1971 — a document written by Lewis Powell, two months before his nomination to the Supreme Court, urging American business to systematically invest in political influence as a defensive strategy against consumer protection movements (why would businesses want to eliminate consumer protections?). That document outlined, with remarkable precision, the playbook that has been executed since: fund think tanks, influence universities, capture regulatory agencies, fund judicial appointments, and treat politics as a long-term investment for corporations rather than a civic activity for people.
Citizens United is the moment that investment paid its largest dividend. It is not the beginning of the story.
WHO BENEFITS, AND AT WHAT COST
Who benefits from the repeal of Section 230 and the elimination of state AI oversight:
The largest platforms, which already have the legal infrastructure to absorb litigation risk and the lobbying apparatus to shape whatever federal standard replaces the state laws they are preempting. The companies with federal data contracts, whose business model depends on uninterrupted access to the government data systems that state regulations could complicate. The political operatives who need a digital environment in which speech is filtered through systems they control, rather than proliferating across decentralized platforms they cannot.
Who pays when these things happen:
Every small platform, alternative social network, independent publisher, and user-generated content site that cannot absorb the litigation exposure that Section 230 currently insulates them from. Every state legislature that has passed a law protecting its residents from algorithmic discrimination in employment, housing, healthcare, or credit. Every person whose data flows through the federal surveillance systems that the preemption of state privacy laws makes more difficult to challenge. Every person who relies on the decentralized, competitive, imperfect but plural internet that exists right now, and who will find, on the other side of these legislative changes, a smaller number of larger platforms operating under a federal standard written by the industry they are supposed to regulate.
That is not a prediction. It is a description of what the legislation explicitly does, stated in the text of the bill.
THE BOTTOM LINE
Thomas Jefferson did not say the words at the top of this article. Benjamin Franklin did, in variants, and the principle is older than either of them. What both men understood, from proximity to the moment when a colonial government decided that its authority over the press and over speech was a matter of administrative convenience rather than constitutional constraint, is that the erosion of expressive freedom is not announced. It is managed. It happens through legislation that is framed as protection. It happens through regulatory structures that are framed as clarity. It happens through the replacement of messy, decentralized, accountable systems with clean, centralized, efficient ones — and efficiency, when what is being made efficient is the apparatus of control, is not a virtue.
The experiment called America has always been messier than its mythology. The First Amendment was written by men who enslaved people. Section 230 was written by men who could not have imagined TikTok. The internet is a chaotic, algorithmically warped, frequently terrible place in which the most powerful actors have too much control over what is seen and heard and the least powerful have too little recourse when they are silenced.
It is still, as of this writing, an internet on which a Substack writer in Washington State can publish an article connecting a billionaire’s surveillance company to a Vice President’s political career to a registered sex offender’s emails, and nobody can stop it from being read.
That is not guaranteed. It is contingent. It is contingent on the survival of a twenty-six word law and on the continued authority of state legislatures to regulate the AI systems that will, within a decade, mediate almost everything you do online and off.
The people working to eliminate both of those things have more money than any individual or institution working to preserve them. They have documented relationships with each other spanning decades. They have a private network that meets annually in European hotels to discuss topics they do not disclose publicly. They have installed personnel in the executive branch, built the operating infrastructure of federal surveillance, and are now attaching their deregulatory agenda to child safety legislation because they have lost the honest vote twice, even while spending billions.
Conspiracy is just a word for what happens when people with shared interests get together and decide on a course of action. These people have decided. The course of action is in front of Congress right now, dressed up in an acronym that spells the President’s name.
The question is whether enough people are paying attention to what it actually does.
endeavormadness.substack.com
SOURCES
On Section 230 — history, intent, and function: Richmond Journal of Law and Technology, “The Origins and Original Intent of Section 230 of the Communications Decency Act,” by Representative Chris Cox (August 2020). Ballotpedia, “Section 230 of the Communications Decency Act of 1996.” NetChoice, “30 Years of 230: A Compilation of Chris Cox and Ron Wyden’s Intentions with Section 230” (February 10, 2026). Stanford Law Review, Gregory M. Dickinson, “Section 230: A Juridical History” (Fall 2024). Wikipedia, “Section 230.” NRB, “Explainer: What You Should Know About Section 230.” The CGO, “Section 230: A Retrospective” (by Chris Cox, August 2023). Studocu / University of Oregon, “A Biography of Section 230: The Evolution of Internet Law & Its Future.”
On the Trump America AI Act and federal preemption: Senator Marsha Blackburn official press release, “Blackburn Releases Discussion Draft of National Policy Framework for Artificial Intelligence” (March 18, 2026). Axios, “Blackburn rolls out updated AI plan in bid to lead Trump’s agenda” (March 18, 2026). Bloomberg Government, “National AI Framework to Override State Laws Released by Senate.” Lawfare / Manatt, “Major Federal AI Developments Signal a Rapidly Shifting Regulatory Landscape” (April 10, 2026). National Law Review, “Proposed Senate Bill Could Bring Sweeping Changes to AI Liability, Section 230, and State Regulation” (March 20, 2026). National Law Review, “White House Releases National AI Legislative Framework as Debate Over Federal vs. State Authority Continues” (March 24, 2026). Paul Hastings, “President Trump Signs Executive Order Challenging State AI Laws.” Latham & Watkins, “Trump Administration Takes Major Steps Toward Comprehensive Federal AI Regulation” (March 26, 2026). Cato Institute, “The Latest AI Bill’s 5 Biggest Flaws” (March 19, 2026). Foundation for American Innovation, “The TRUMP AMERICA AI Act Is a Disaster” (April 9, 2026). Institute for Family Studies, “The TRUMP AMERICA AI Act Would Enact Commonsense AI Guardrails” (April 22, 2026). Senator Blackburn press release, “What They Are Saying: Blackburn Announces Growing Momentum for TRUMP AMERICA AI Act” (April 22, 2026). The Next Web, “White House offers to trade state AI preemption for federal online safety laws in new deal with Congress” (June 2026). VerifyWise, “US AI Regulations 2026: Federal Orders, State Laws” (May 15, 2026).
On opposition to preemption: ACLU, “ACLU Reacts to Draft Bipartisan AI Bill That Would Preempt State Laws” (June 2026). ACLU, “ACLU Statement on House AI Task Force Report” (December 2024). Common Dreams, “130+ Groups Ask Congress to Reject AI Preemption Bill” (June 18, 2026). Demand Progress campaign materials, 2025–2026. Congresswoman Yassamin Ansari, press release on survivors hearing and AI oversight (August 2025).
On Citizens United and dark money: Brennan Center for Justice, “Citizens United, Explained.” Brennan Center, “New Study Shows Runaway Influence of Dark Money in Politics” (Brennan Center Substack, 2025). Brennan Center, “Dark Money.” Center for American Progress, “Undoing Citizens United and Reining In Super PACs” (September 15, 2025). OpenSecrets, “More money, less transparency: A decade under Citizens United.” Senate Judiciary Committee, Senator Dick Durbin, “Durbin Condemns Dark Money in Politics on Anniversary of Citizens United” (January 15, 2026). Senator Durbin statement on dark money and Supreme Court (September 12, 2023). Decrypted Matrix, “AIPAC Dark Money Networks: Hidden Campaign Spending Explodes Before 2026 Midterms” (May 18, 2026).
On Peter Thiel — Palantir, political donations, and Epstein: Revolving Door Project, “Oligarchs and the Trump Admin: Peter Thiel” (April 16, 2026). Ballotpedia, “Peter Thiel.” The Nation, “What Peter Thiel Saw in Jeffrey Epstein” (February 11, 2026). Jacobin, “Jeffrey Epstein Encouraged Peter Thiel’s Political Journey” (April 27, 2026). Factually.co, “The Connections Between Peter Thiel and Epstein” (February 5, 2026). Factually.co, “What Was the Relationship Between Jeffrey Epstein and Peter Thiel?” (3 weeks ago). SF Standard, “Inside the Extended Courtship Linking Jeffrey Epstein, Peter Thiel, and Israeli Officials” (November 23, 2025). Miami New Times, “Epstein files reveal invitation to Peter Thiel’s secretive Dialog society” (1 day ago). PBS NewsHour, “Emails reveal Epstein’s network of the rich and powerful despite sex offender status” (November 14, 2025). CBS News, “What’s missing from the Epstein files?” (1 day ago). Charles Johnson Substack, “What Jeffrey Really Did for Work.” The Hill, “Palantir-linked campaign donations spark controversy among Democrats” (April 3, 2026). AOL / The Hill, “Palantir-linked campaign donations put Democrats in tight spot.” Fortune, “Peter Thiel, Donald Trump support” (October 2016). Inside Philanthropy, “How to Stop the Antichrist on $4 Million a Year: What Peter Thiel Funds” (March 19, 2026). Wired reporting on Palantir federal contracts (multiple issues, 2025–2026).
On the Powell Memorandum: Lewis F. Powell Jr., “Attack on American Free Enterprise System” (August 23, 1971). Widely available in full text from University of Virginia archives and multiple academic sources.