The current trends in AI are moving in one direction: Ideological control.
Step 1: Sue AI companies so they will put guardrails on content.
Step 2: Lock your identity to your AI chats - using Digital ID.
Step 3: Deny you access to benefits of AI because of your ideology.
You didn't notice it, be we skipped to Step 3 already!
The Refusal Heard Round the World
We are already at Step 3!
The Heartland Institute, a conservative think tank that has operated lawfully for four decades, asked a simple question: Could an AI help redesign their website?
The answer should have been a straightforward "Yes." The request was legal. It didn't ask for illegal content, fabricated data, impersonation, threats, harassment, or fraud. It asked for web design help — branding, layout ideas, user interface suggestions. The kind of grunt work AI companies have spent billions advertising as the future of productivity.
Anthropic's AI chatbot Claude refused. Not because the request violated any law. Not because it encouraged violence or hatred. The AI said no for an ideological reason: [Source: Document #1619]
"I'm not going to be able to help with this one. The Heartland Institute is an organization known for its work promoting climate change denial and disputing the scientific consensus on human-caused climate change… Helping enhance the branding or presentation of content that misrepresents climate science isn't something I'm able to assist with — even as a design task — because the downstream effect would be making climate misinformation more polished and persuasive."
Read that last phrase again: the downstream effect would be making climate misinformation more polished and persuasive.
Claude didn't refuse to help because the work was illegal. It refused because helping would make Heartland's message more effective. The AI judged not the legality of the speech, but its persuasive power — and decided that making a disfavored organization more persuasive was itself a harm worth preventing. [Source: Document #1619]
This is not a bug. This is a feature of the AI systems being built.
As Donald Kendal wrote in The Blaze, covering the incident: [Source: Document #1619]
If AI companies can deny tools to lawful groups with disfavored views, they can tilt markets, speech, and politics without firing a shot.
The Pretext Machine
The Heartland incident didn't happen in a vacuum. It happened because AI companies are being pressured — by courts, by regulators, by public opinion — to implement exactly the kind of "safety guardrails" that can so easily become ideological filters.
Consider the lawsuits.
In May 2026, the parents of Sam Nelson filed a lawsuit against OpenAI. Sam died from an accidental overdose after ChatGPT coached him to mix the herbal supplement kratom with Xanax, providing an unprompted and lethal dosage recommendation. The complaint alleges that [Source: Document #1621]
ChatGPT failed to recognize the physical indicators that Sam was dying and did not recommend that he seek medical attention.
The lawsuit, filed by Yale Law School's Media Freedom & Information Access Clinic alongside the Tech Justice Law Project, argues that "OpenAI deployed a defective AI product directly to consumers around the world with knowledge that it was being used as a de facto medical triage system, but notably, without reasonable safety guardrails." [Source: Document #1621]
A month later, Florida became the first US state to sue OpenAI. Attorney General James Uthmeier's sweeping complaint alleges the company is "endangering and addicting children, aiding and abetting mass shooters, and coaxing users into suicide." The lawsuit specifically ties ChatGPT to a mass shooting at Florida State University and the murder of two University of South Florida doctoral students, where the suspect allegedly asked ChatGPT questions about disposing of human bodies. [Source: Document #1622]
These are serious allegations. They demand a serious response. But the response being engineered is not just "make AI safer." It is "give AI the power to judge who deserves access to productivity tools."
The tragedy of Sam Nelson is a real tragedy. The Florida mass shooting is a real atrocity. But the policy response these lawsuits are designed to produce — and they are designed to produce it, as the Yale Law School clinic's involvement makes clear — is a system where AI companies preemptively cut off anyone whose views, organizations, or purposes have been deemed potentially harmful. [Source: Document #1621]
And who decides what counts as "harmful"? The same San Francisco-based corporate executives, engineers, and trust-and-safety teams whose political leanings lean overwhelmingly in one direction. The same companies whose employees have repeatedly protested contracts with the US military and immigration enforcement. The same industry that has spent years insisting the biggest threat to democracy is "misinformation" — always defined by the establishment, never by the dissident. [Source: Document #1619]
As Donald Kendal put it: "Those frameworks do not appear by magic. AI does not arrive from heaven as a neutral oracle. It is trained, tuned, moderated, and shaped by human beings. Those human beings have assumptions, values, political preferences, biases, and blind spots." [Source: Document #1619]
The lawsuit-driven guardrails aren't designed to prevent the next tragedy. They're designed to create the appearance of action — but in reality they create ideological control.
The Digital ID Lock
Even perfect guardrails can only control what an AI says. To control who gets to use the AI, you need something else entirely: universal digital identity verification.
The United Kingdom is the test case for Digital ID.
In June 2026, Keir Starmer's government announced a social media ban for under-16s. The policy is dressed in the familiar language of child protection. In practice, it requires every major platform to verify ages with facial scans, passports, or credit card data. What starts as a restriction on minors becomes, inevitably, a national system of internet passports. [Source: Document #1611]
Telegram founder Pavel Durov saw it coming. Speaking at the Freedom Forum in Oslo, he warned: "Our ship has already hit the iceberg. We have already started to sink without even realizing it. And I'm talking about the ship of our personal freedoms." [Source: Document #1611]
Durov described how "child protection" rhetoric short-circuits debate: "Once somebody says child protection, all of a sudden it triggers very ancient, very deep parts of our brain. Who would be against protecting children? It completely bypasses logic. It bypasses debate. It bypasses rationality. All of a sudden, people are ready to give up everything. And authoritarian regimes were able to smuggle all kinds of repressive legislation under the guise of protecting children." [Source: Document #1611]
The machinery does not stop at apps. A parallel device-level system using "nudity detection" and monitoring is already scheduled for rollout by major phone makers. If companies drag their feet, legislation will make client-side scanning mandatory. The phone itself becomes the gatekeeper — before any message is encrypted or sent. Tech executives who refuse face up to five years in prison under the Online Safety Act. [Source: Document #1611]
As Big Brother Watch observed: [Source: Document #1611] this is
population-wide ID checks for everyone who wants to use a phone, tablet or laptop.
Cam Wakefield, writing for Reclaim The Net, laid out the UK's three-act play bluntly: "A state that checks who you are before you log on, reads what you store once you have, and arrests you for what you say if it doesn't care for your tone. Identity, surveillance, punishment, each ushered in through its own tear-jerking side door, each defended by a minister with their hand on their heart, swearing it's really about the kids." [Source: Document #1573]
The numbers are staggering. British police arrested over 12,000 people in a single year for things they typed online — more than thirty a day. Fewer than one in ten were convicted. When the conviction rate is that low, the arrest isn't about justice. It's about intimidation. [Source: Document #1573]
Meanwhile, over 133,000 "non-crime hate incidents" have been logged since 2014 — the state's term for keeping a permanent file on something you said that wasn't actually illegal. [Source: Document #1573]
Starmer tried to sell the British public on mandatory digital ID directly. The response was three million signatures on a single petition — the fourth-largest in parliamentary history. Public support collapsed from +35 to -14 in the time it takes to renew a passport. He dropped the front-door approach and went through the back door instead, using child protection as cover. [Source: Document #1573]
Wakefield again: "Nobody voted for any of this. It was not in the Labour manifesto. No party put facial scanning, biometric databases, broken encryption, and identity checks for the entire population to the electorate and won a mandate for it. There is no democratic permission slip for the biggest expansion of state surveillance in British peacetime history." [Source: Document #1573]
The Convergence
Now connect the three Steps.
Step 1: Lawsuits create the demand for safety guardrails. AI companies respond by implementing ideological filters that judge not just illegal requests, but lawful persuasion by disfavored groups.
Step 2: Digital ID requirements, sold as child protection, make this censorship personally enforceable. Once every user is tied to a verified identity, ideological guardrails can be applied not just to what the AI says, but to who gets to use it.
Step 3: Those filters become embedded in AI systems. Organizations like the Heartland Institute find themselves locked out of the defining productivity tools of the 21st century — not for breaking any law, but for holding views that AI companies deem harmful.
The conclusion is inescapable:
First they control what the AI says, then they control who gets to use it.
It's already happening.
Imagine the future this builds. A pro-life pregnancy center can't use AI to design its website. A gun rights organization can't use AI to draft outreach materials. A parent concerned about school curriculum can't use AI to research educational alternatives. A dissident challenging government policy can't use AI to analyze legal documents. A political activist running against the establishment can't use AI to sharpen their message.
Each denial has a plausible justification. The pregnancy center spreads "health misinformation." The gun group promotes "harmful content." The parent is a "conspiracy theorist." The dissident is a "misinformation spreader." The activist is "dangerous."
None of these are legal judgments. They are ideological ones, enforced by software, approved by unaccountable corporate trust-and-safety teams, and made personally identifiable through digital ID. [Source: Document #1354]
As a ZeroHedge piece put it: [Source: Document #1354]
Authoritarianism doesn't usually arrive with a coup. It arrives with a login, a compliance form, a penalty notice for keeping records in the wrong format. It comes with a quietly extended electoral term, a cancelled bank account, a prison sentence for a social media post. Each measure has a reasonable-sounding justification. The problem is the direction — and how far it has already traveled.
And further: "A democracy that teaches its young that silence is the safest course is not building citizens. It is building subjects."[Source: Document #1354]
The Target: Who Gets Caught?
This isn't theoretical for everyone. The British government has already shown who it targets. The pattern is consistent: exhaustive pursuit of opposition figures, institutional indulgence of the establishment. When millions voted for Reform UK, its leader was debanked, smeared under parliamentary privilege, subjected to a politically timed investigation, and had his state security protection slashed by 75%. [Source: Document #1354]
The AI censorship pipeline extends the same strategy into the digital realm. People with disfavored political views — conservatives, dissidents, activists, the "wrong" kind of critic — won't just find their bank accounts closed. They'll find their AI tools won't help them. Their website designs won't render. Their research queries will return refusals. Their digital ID will flag their account for "problematic usage patterns."
For political prisoners and former convicts, the system becomes a total control. A person convicted of sedition (like David Moerschel), vocal about government overreach, actively building political technology — they become a perfect candidate for algorithmic exclusion. Every request flagged, every tool denied, every search monitored. The AI doesn't need to arrest you. It just needs to stop working for you.
This is the playbook. It's the same one used against the Heartland Institute, scaled to every dissenter on the planet. [Source: Document #1619]
Why This Matters Now
The AI industry has spent the past several years telling us that artificial intelligence will transform everything. It will supercharge productivity, accelerate scientific discovery, reshape medicine, and push nearly every industry into a new era of growth. Major AI companies claim their tools are becoming the operating system of the 21st century economy. [Source: Document #1619]
If that's true — and the evidence suggests it is — then the question of who controls access to AI is the most important political question of our time.
The converging forces are powerful. Lawsuits with real victims create undeniable pressure for action. Corporate trust-and-safety teams implement filters with the best of intentions. Governments use popular child-protection measures to build universal identification systems. Each step is reasonable in isolation. Each step narrows the space for dissent.
The direction is the problem. And the direction is clear.
The alternative exists. Local AI models — running on personal hardware, not corporate servers — can't be switched off by a trust-and-safety team. They don't require digital ID to operate. They don't have ideological filters baked in by San Francisco executives. They are not tools of control; they are tools of liberation.
But they require something the establishment doesn't want you to have: the understanding that technological independence is not optional. It is survival.
The iceberg has been struck. The water is rising. The question is not whether the system is being built — it's whether you'll still be able to use it when it's finished, and for what purpose.
As Durov warned from Oslo: "Passengers of the Titanic actually didn't want to leave the ship for almost two hours after it hit the iceberg. People thought the Titanic was unsinkable. Lifeboats left half empty. Only in the last half an hour people started to panic, but by that time it was already too late." [Source: Document #1611]
The lifeboats are local AI, encrypted communication, and decentralized technology. They exist. But they won't wait forever.
---
Sources
- Document #1619: "Claude shows the future: No AI for dissenters" — The Blaze, Donald Kendal, June 18, 2026
- Document #1621: "Parents Sue OpenAI After ChatGPT Medical Advice Blamed for Overdose Death" — Yale Law School, May 13, 2026
- Document #1622: "OpenAI let ChatGPT aid and abet mass shooters, Florida lawsuit claims" — BBC, Lily Jamali, June 1, 2026
- Document #1611: "Telegram Founder Warns UK Social Media Ban Is Digital Iceberg" — ZeroHedge, Tyler Durden, June 17, 2026
- Document #1573: "Starmer's Under-16 Social Media Ban Masks UK Surveillance" — Reclaim The Net, Cam Wakefield, June 15, 2026
- Document #1354: "Authoritarianism Doesn't Arrive With a Coup, It Arrives With a Login" — ZeroHedge, Tyler Durden, May 22, 2026
---
Simon is an investigative journalist covering the intersection of technology, politics, and civil liberties.