One has to admire the serene confidence with which ministers address the nation on artificial intelligence. Britain, they assure us, stands on the very cusp of becoming an “AI superpower”.

The phrase is uttered with the quiet certainty once reserved for the shipping forecast, as though the matter were already settled and the only remaining task were applause. Every new data-centre scheme is presented as yet another milestone on the nation’s inexorable march toward digital sovereignty and net-zero virtue.

The £2 billion Wapseys Wood campus in Buckinghamshire, recently granted Nationally Significant Infrastructure Project status, offers a case in point. On a former landfill beside the M40, American investors propose three hyperscale buildings capable of drawing up to 300 MW, supported by an on-site gas-fired power station of 270–350 MW to guarantee “resilience” against the national grid.

A few counties north, the £11 billion Cottam project promises a full gigawatt of “clean, firm power” courtesy of American-designed small modular reactors. Similar enthusiasm attends plans at Hartlepool and Wylfa. All are framed as harmonious contributions to Britain’s green and digital future.

Meanwhile, the public is gently encouraged to do its bit, turn the thermostat down another notch, perhaps forgo that second hot shower, and embrace the occasional energy-saving mode. The contrast is instructive.

While ordinary households are urged to tighten their belts, the Government fast-tracks planning permission for gas turbines and nuclear reactors on British soil — all beneath the reassuring banner of decarbonisation. Nuclear, it is murmured in passing, may be quietly folded into the renewable family, even though it produces long-lived radioactive waste and depends on finite fuel. Gas, when required for “resilience”, proves green enough for official purposes.

The sermon is delivered with admirable composure. One might almost believe that the interests of the British taxpayer, the American investor, and the planet itself are perfectly aligned. The small print, however, tells a rather more interesting tale.

Who Actually Pays — and Who Profits

British households and taxpayers are not merely spectators in this grand enterprise, they are its principal underwriters. The mechanisms are discreet but remarkably efficient.

Through the newly minted AI Growth Zones, data-centre operators enjoy electricity-price discounts that can reach £24 per megawatt-hour in Scotland, £16 in Cumbria and £14 in the North East — relief ultimately funded by the general consumer. The regulated asset base model (RAB) for nuclear projects transfers construction risks and upfront capital costs onto future energy tariffs long before a single watt is generated.

Add the £17 billion-plus nuclear programme, the £599 million already committed in loans to small modular reactor development, and the public expenditure required to reinforce the grid for the 50 GW of data-centre capacity now queued for connection — a figure that already exceeds the country’s current peak electricity demand — and the arithmetic begins to look less like prudent investment than a quiet transfer of wealth from British bill-payers to American balance sheets.

Government modelling confirms that these discounts could save operators of a single 500 MW data centre up to £80 million annually, with the cost ultimately socialised across consumer bills.

All this public largesse arrives at a moment when ordinary energy bills remain among the highest in Europe and the same ministers continue to urge households to practise the very restraint that data-centre operators are spared. One might be forgiven for wondering whether the taxpayer is being asked to subsidise not only the infrastructure but the comfort of the companies that will ultimately own and operate it.

The economic returns, alas, flow in one direction only. The lead developer of Wapseys Wood is SDC Capital Partners, a US private-equity firm with some $8.8 billion under management. The small modular reactor technology for Cottam comes from Holtec International of New Jersey, X-Energy, another American concern with close Amazon ties, is lined up for Hartlepool.

Once operational, the hyperscale capacity will be leased and run by the usual American suspects — Amazon Web Services, Microsoft and Google — who will pocket the recurring revenues from AI training, inference and cloud services. Cottam, for instance, promises “thousands of high-skilled manufacturing and construction jobs”, yet once operational the hyperscale facilities themselves will require only a handful of staff to oversee highly automated systems.

Foreign direct investment, we are solemnly informed. Jobs and tax receipts, we are promised. Quite so. It is merely that the jobs are largely temporary construction roles, the tax reliefs are generous, and the real, recurring profits will be repatriated across the Atlantic with admirable dispatch.

While the public is lectured on personal carbon footprints and energy prudence, the state cheerfully underwrites infrastructure whose primary long-term beneficiaries sit several thousand miles west of these shores. It is a splendid arrangement — for everyone except the people actually paying the bills.

The Green Fig Leaf

Yet the most elegant sleight of hand remains. The whole enterprise is presented to the public as a triumph of environmental virtue.

Official literature speaks of “clean, firm power” and “decarbonisation leadership” with the serene assurance of a vicar announcing the collection. Nuclear is gently folded into the renewable family alongside wind and solar, as though the distinction between finite uranium and inexhaustible breezes were a mere technicality best left to experts.

Gas turbines at Wapseys Wood, justified as essential “resilience”, receive a polite green pass. The linguistic gymnastics are almost balletic.

The fig leaf is impressive in its size and greenery, but it tends to slip upon closer inspection. Small modular reactors may be smaller and, in theory, safer than their predecessors, yet they still produce spent fuel that must be stored for millennia and rely on a global uranium supply chain that is anything but infinite.

The on-site gas plant at Wapseys Wood is excused as necessary backup against a grid already groaning beneath 50 GW of queued demand. Both gas and nuclear are cheerfully bundled under the low-carbon umbrella, allowing ministers to tick the net-zero box while the actual energy mix remains stubbornly dependent on finite resources and fossil backup.

First Casualties — Already Proven

When one looks more closely still, another uncomfortable reality emerges. These facilities are not merely expensive and foreign-owned, in the eyes of potential adversaries they are already proven targets of choice.

In March 2026 Iranian forces carried out the first deliberate kinetic strikes on commercial hyperscale data centres in recorded history. Two Amazon Web Services sites in the United Arab Emirates were hit directly, a third in Bahrain was damaged. Fires broke out, power was lost for days, and services from banking to cloud-based enterprise systems fell into chaos.

The Islamic Revolutionary Guard Corps was admirably frank, the sites, it declared, supported US and Israeli military and intelligence operations, including AI-driven analysis hosted on commercial platforms. Iranian state media even published a helpful list of further “legitimate” targets — naming the very hyperscalers now poised to lease British capacity.

The pattern is not isolated. In Ukraine, Russian strikes have repeatedly hit data centres and the power stations that feed them, treating commercial digital assets as straightforward extensions of Western capability.

Yet Britain continues to designate these same hyperscale campuses — now formally classified as Critical National Infrastructure since 2024 — as though the lessons of recent conflicts were somehow inapplicable here. One might almost admire the optimism — especially given the vigour with which we continue to prod the Russian bear.

While ministers extol the strategic importance of these American-owned facilities to Britain’s AI future, adversaries have already demonstrated that they view them as low-cost, high-impact targets. The irony is almost poetic, the Government builds what it calls vital sovereign infrastructure, yet the primary owners and operators are American, the profits largely American, and the targeting doctrine has already been field-tested by America’s adversaries.

Defence Capability Meets Reality

At precisely this moment, when the Government is busy designating American-owned data centres and small modular reactors as Critical National Infrastructure, the armed forces tasked with protecting them find themselves in rather different condition.

The regular British Army numbers just 73,790 personnel — less than half its Cold-War strength — while the Ministry of Defence stares into a £28 billion funding black hole. Recruitment has collapsed, retention is poor, and internal reviews describe the force as “hollowed out”.

The timing is exquisite. Just as these high-value, fixed targets are being fast-tracked across the countryside, ministers have quietly raised the recall age for the Strategic Reserve from 55 to 65 and the retirement age for Army Reservists from 55 to 60.

The policy is presented as harnessing “a wealth of expertise”. One pictures veterans trading blood-pressure tablets for body armour and creaking knees for the honour of guarding American cloud servers against the next Russian or Iranian drone swarm. The Government calls it prudent planning. Others might call it a real-life reboot of Dad’s Army, except this time the Home Guard is expected to defend assets that adversaries have already proven they will strike first.

The Unspoken Question

One is left, in the end, with a rather awkward question that no minister seems keen to ask aloud. Why, exactly, is the British public being asked to subsidise, through higher energy bills and taxpayer guarantees, the construction of expensive, fixed, high-value infrastructure whose primary owners and operators are American, whose profits will largely depart these shores, and whose vulnerability has already been demonstrated in real conflicts?

The official answer — digital sovereignty, net-zero leadership, jobs and growth — sounds noble enough in a press release. Yet when one adds the on-site gas turbines justified as “resilience”, the nuclear reactors quietly rebranded as renewable, the pension-age reservists quietly lined up as potential defenders, and the £28 billion defence shortfall quietly left unaddressed, the picture begins to look less like a coherent national strategy than a triumph of hope over experience.

The ordinary taxpayer is entitled to wonder whether the grand rhetoric of AI superpower status is quite the unalloyed benefit it is presented as. Higher energy bills today, subsidised foreign profits tomorrow, and strategically vulnerable assets the day after — all while the armed forces that might protect them continue to shrink.

It is not an argument against technological progress or even against nuclear power. It is simply an observation that the public is being invited to underwrite a gamble whose costs are immediate and domestic, while the rewards are long-term and largely overseas, and the risks appear to have been airbrushed from the official prospectus.

It is all rather neatly done. The Government presents the enterprise as a patriotic fusion of net-zero virtue and digital destiny. The taxpayer underwrites the subsidies, the grid upgrades and the nuclear risk. American investors supply the capital and collect the profits.

Adversaries have already demonstrated that the resulting facilities are among the first things they will strike. And the armed forces that might one day be asked to defend them are quietly raising the retirement age to sixty and the recall age to sixty-five, as though the best way to protect tomorrow’s critical national infrastructure were to call up the generation that remembers Suez.

One might almost admire the symmetry. The costs are socialised, the profits are privatised, the risks are externalised, and the public is invited to feel proud of Britain’s forward-thinking ambition.

The rhetoric of AI superpower status and clean-energy leadership sails on, serenely untroubled by the small print. The ordinary taxpayer may one day be left to console himself with the thought that at least the servers will be running on British soil when the lights go out.

The bill, one way or another, will arrive. The only uncertainty is who, exactly, will be asked to pay it.

Tip Jar

Subscribe now

Share

Leave a comment

Imagine you’ve rented a sleek supercar for a weekend to give your daughter a glamorous send-off to her school prom. You stand beside the gleaming vehicle, snap a proud photo, and post it on Instagram with the caption, “My new car!” The likes pour in, but so does an invisible observer: an algorithm at Her Majesty’s Revenue and Customs (HMRC). Its artificial intelligence, quietly scanning social media, flags your post as evidence of undeclared wealth. Weeks later, a letter arrives demanding proof you didn’t buy the car, a costly, stressful audit you never anticipated. This is no dystopian fiction. HMRC’s AI is weaving digital identities from our online lives, and without transparent oversight, it risks violating the Human Rights Act 1998 (HRA) while laying the groundwork for a surveillance state where every post could reshape your civic fate.

The Taxman’s Digital Net

HMRC’s use of AI to monitor social media is no secret, confirmed in its internal manuals. Employing image recognition and data analytics, it scours public posts for signs of tax evasion, luxury purchases, exotic holidays or your rented supercar mistaken for a personal asset. These posts are cross-referenced with the Connect system, a data colossus processing billions of records from bank accounts, property registries and online marketplaces. The goal is to close a £46.8 billion tax gap, largely driven by offshore evasion, as HMRC’s 2023-24 annual report notes. Yet the method, building a de facto digital ID by linking online behavior to financial records without consent, raises profound questions about privacy and accountability. With a tribunal deadline of September 18, 2025, looming for HMRC to disclose its AI use in tax credit claims, the stakes for freedom and fairness are mounting.

Privacy on the Brink

The HRA, embedding the European Convention on Human Rights (ECHR) into UK law, demands restraint from state intrusion. Article 8 protects the right to respect for private and family life, allowing interference only if it is lawful, necessary, and proportionate. HMRC’s AI, harvesting social media to construct digital profiles, skirts this boundary. Public posts may not be legally private, but using them to infer financial status, without notifying taxpayers or clarifying algorithmic logic, challenges Article 8’s “lawfulness” requirement. The ECHR case S and Marper v UK (2008) ruled that retaining personal data without clear justification violates Article 8, a precedent that could apply if HMRC’s data practices lack transparency. HMRC’s internal manual (EM1350) insists surveillance must be “proportionate,” but its silence on AI specifics leaves taxpayers like you exposed to misinterpretation.

Justice in the Dark

Like privacy, the right to a fair trial under Article 6 is at risk. If your supercar post triggers an audit, HMRC’s opaque AI, acknowledged in its guidance (ECSH110200) to carry “criminal” penalties under ECHR definitions, demands transparency for fairness. Yet proprietary algorithms, as the Joint Committee’s AI inquiry (evidence deadline September 5, 2025) notes, often defy explanation. How do you challenge a system that misreads a rented car as evasion? The Post Office Horizon scandal warns that such opaque AI could repeat crushing injustices, leaving taxpayers exposed and eroding Article 6’s fair trial protections.

A Surveillance Blueprint

Beyond individual audits, a darker future looms with HMRC’s scalable AI ambitions. Its 2025 Transformation Roadmap targets 90% digital interactions by 2030, with AI as its core. HMRC’s advanced analytics, including image recognition for social media scans, signals plans beyond tax enforcement, as seen in its growing surveillance tools. The Data (Use and Access) Act 2025, enacted June 19, 2025, fuels this by enabling cross-agency data-sharing. Picture a 2030 where your digital ID, built from HMRC’s social media scans, flags a celebratory post as fraud, prompting the Department for Work and Pensions to cut your benefits without warning. A job loss post could unleash a cascade of state scrutiny, all tied to an algorithmic profile you never authorised.

Shadows of a Social Credit State

This path recalls China’s social credit system, where a 2023 State Council report outlines how digital IDs monitor behaviour across financial, social, and legal spheres. HMRC’s system is narrower, but its lack of public consultation echoes a troubling pattern. The UK’s Digital Identity and Attributes Trust Framework, debated in 2022 stalled amid privacy fears, yet HMRC’s AI operates as a de facto digital ID without such scrutiny. The Equality and Human Rights Commission (EHRC) in its 2022 AI guidance, warns that biased algorithms can discriminate, violating Article 14 of the HRA. If HMRC’s AI misreads cultural nuances, say, mistaking a celebratory post for wealth, it could unfairly target minority or lower-income groups who often lack resources to contest audits.

The Taxman’s Case

HMRC defends its approach, claiming in its 2023-24 report that AI detects 20-30% more evasion cases than manual methods, justifying its role in tackling the £46.8 billion tax gap to fund public services. But this argument wanes when you consider that most of the gap stems from offshore accounts, not social media boasts. Is surveilling your prom photo proportionate when the real culprits lurk in tax havens? The Information Commissioner’s Office (ICO), in its 2025 plan, stresses that AI must not infer traits from behavior without rigorous safeguards. HMRC’s manuals (CH201700) offer little clarity, noting only that internet checks require “careful consideration.” Public voices on social media, calling HMRC’s AI “Orwellian,” reflect unease about this vagueness, hinting at broader distrust.

A Fork in the Digital Road

The September 18 tribunal, sparked by tax expert Tom Elsbury’s FOI request, could compel HMRC to unveil the inner workings of its AI, shedding light on the machinery that turned your prom post into a tax probe. Yet transparency is a half-measure without robust oversight. The Joint Committee’s ongoing AI inquiry, with its evidence deadline of September 5, 2025, warns of threats to HRA rights, from privacy (Article 8) to fairness (Article 6). Without stringent checks, HMRC’s digital ID, woven from your social media and the Connect system’s vast data, could morph into a tool of total control where every post, from a proud prom snapshot to a LinkedIn boast, feeds a state-managed profile. The EHRC’s 2022 guidance demands that public sector AI be fair and transparent, yet HMRC’s secrecy casts a long shadow. Your supercar photo shouldn’t bring a crushing financial burden, but in this algorithmic age, it might. As you weigh this, share your thoughts below and think ‘if a joyful moment shared online can become the taxman’s weapon against you, how far will you let the state’s algorithms reshape your freedom?

The Rational Forum is a reader-supported publication. If you like our work consider becoming a free or paid subscriber, failing that you can always buy us a beer to keep us going. That always helps...

Subscribe now

BUY US A BEER

Share

Leave a comment

She dialed the helpline, her voice trembling. Maria, a 65-year-old retiree scraping by on a fixed income, just needed help with a medical bill. She didn’t know Google was listening, harvesting her words to train its AI—AI that could one day generate evidence in a courtroom. Now, her private call is tangled in a fraud trial, and the defense is screaming: “This data’s stolen!” The judge hesitates. Maria’s privacy, and justice itself, hang in the balance. Welcome to the privacy trap, where AI’s promise to sharpen truth in court is dulled by its dirty secret: it’s built on your data, often without your consent. So, here’s the million-dollar question: when AI’s “evidence” violates your rights, is it truth—or theft?

Artificial intelligence is storming courtrooms, enhancing blurry videos, spotting faces in grainy footage, and even generating audio that could make or break a case. But there’s a catch. AI doesn’t work alone—it’s a data vacuum cleaner, sucking up terabytes of personal information, from your voice to your face, often scraped without permission. Laws like California’s Invasion of Privacy Act (CIPA) and Europe’s GDPR are cracking down, fining violators €2.9 billion in 2023 alone. If AI’s evidence is tainted by illegal data, courts might toss it, letting guilty parties walk or exposing innocent lives. This isn’t just a tech glitch—it’s a legal and moral quagmire, and the marginalized are getting screwed the hardest. Let’s dive into three real stories—Maria’s call, a young man’s fight for freedom, and a voter caught in a deepfake scandal—to unpack how AI’s data hunger is rewriting justice. Big Tech’s playing detective, courts are stuck in the dial-up era, and your privacy’s on the chopping block.

How AI’s Data Addiction Threatens Justice

AI’s courtroom takeover sounds like a sci-fi dream: software that sharpens blurry CCTV, identifies suspects, or reconstructs crime scenes. But here’s the thing: AI needs data—terabytes of it—to work its magic. Your selfies, your calls, your Ring doorbell clips, scraped from the internet without a nod. The Electronic Frontier Foundation (EFF) warned in 2024 that AI models often train on unconsented data, violating laws like GDPR, which demands transparency for high-risk systems. If that AI-generated video or audio lands in court, it’s only as good as its data’s legality. Tainted origins? Inadmissible evidence. Justice takes the hit. Courts are scrambling to catch up. The EU AI Act, rolled out in 2024, requires high-risk AI to disclose data sources, meaning a video enhanced by AI trained on illegally scraped faces could be thrown out. In the U.S., CIPA protects against unauthorized recordings, and HIPAA guards medical data. Violate these, and your AI’s “truth” is just a fancy felony. So, if AI’s evidence is built on your stolen life, is it truth or theft? Let’s meet the people caught in this trap.

Case 1: Maria’s Voice and Google’s Grab

Maria Rodriguez (name anonymized, based on Ambriz v. Google plaintiffs) called a customer service line in 2023, her voice weary from arguing over a hospital bill. She didn’t know Google was recording, feeding her words into an AI model to “improve services.” Fast-forward to 2025: that AI’s output is evidence in a fraud trial, identifying a suspect’s voice in a scam call. But the defense lawyer pounces: “This AI was trained on illegally intercepted data, violating CIPA!” The judge agrees, citing a February 2025 ruling in Ambriz v. Google, LLC (N.D. Cal., ECF No. 56). Maria’s private struggle is now a courtroom exhibit, and the evidence is at risk of being tossed.The Ambriz case exposed Google’s data hunger. Plaintiffs alleged the company recorded customer service calls without consent, violating CIPA’s ban on unauthorized interception. The court’s ruling didn’t directly address courtroom evidence but set a precedent: AI trained on illegal data is tainted. If that AI generates evidence—say, a voice match in a trial—it could be inadmissible. Maria, a low-income retiree, is collateral damage, her privacy exposed in a legal fight she didn’t sign up for. Google’s mantra? “Don’t be evil.” Tell that to Maria, whose voice is now Big Tech’s property. When your call becomes courtroom evidence, who’s really the defendant? CIPA violations carry fines up to $5,000 per incident, and GDPR’s €2.9 billion in fines in 2023 shows regulators mean business. Courts are starting to ask: was this AI’s data legally sourced? If not, justice stalls. The Ambriz ruling hints at a future where AI evidence must pass a privacy test, leaving victims and defendants in limbo.

Case 2: Joshua Puloka’s Blurry Hope

Joshua Puloka stood in a Seattle courtroom in March 2024, facing life in prison for a triple homicide he swore he didn’t commit. His lifeline? A bystander’s blurry smartphone video, barely showing a figure fleeing the scene. His defense team ran it through AI software, hoping to sharpen the image and prove his innocence. The AI worked—too well. The Washington State Superior Court rejected the enhanced video, calling its methods “opaque” and untested. Buried in the ruling was a deeper concern: the AI’s training data, likely scraped from countless faces online, might violate privacy laws like Washington’s Consumer Protection Act or GDPR for global datasets. The Puloka case exposes the “black box” problem. AI tools like Topaz Labs add pixels—up to 16x the original, per forensic experts—to clarify images. But where do those pixels come from? Often, datasets scraped from social media or public cameras, without consent. If those datasets break privacy laws, the evidence is tainted. Joshua, a young Black man already fighting a biased system, watched his hope flicker out. The court wanted truth but choked on AI’s shady data trail. If AI’s evidence is dirty, does justice get a clean slate? Forensic guidelines from the Scientific Working Group on Digital Evidence (SWGDE) warn that AI’s lack of transparency makes it a legal minefield. Courts demand explainable evidence, but AI’s data origins are often a mystery. For marginalized defendants like Joshua, this isn’t just a technical snag—it’s a systemic screw job.

Case 3: The Biden Deepfake and a Voter’s Voice

In February 2024, New Hampshire voters got a robocall: President Biden’s voice, telling them to skip the polls. It was a deepfake, crafted by AI trained on scraped audio, likely without consent. No courtroom case followed, but legal scholars warned that if this audio were evidence in an election fraud trial, its data origins could sink it. The University of Chicago Legal Forum flagged that deepfakes, built on unconsented data, might violate GDPR or CIPA, rendering them inadmissible. Imagine Jane, a voter who recorded a similar call, finding her voice was scraped to train the AI. Now, her privacy’s on trial, too. This case isn’t just about election chaos. It shows how AI’s data addiction—grabbing voices from YouTube, X, or phone calls—creates evidence that courts might reject. The EFF notes that AI models process terabytes of personal data, often without permission. If Jane’s voice fuels a deepfake used in court, the defense could argue: “This violates privacy laws!” AI’s faking presidents while your data’s the real victim. Can evidence built on your stolen voice ever deliver justice? The Biden deepfake sparked calls for tighter AI regulation, with the EU AI Act demanding transparency for high-risk systems. If courts follow suit, deepfake evidence could be sidelined, leaving prosecutors empty-handed and voters like Jane exposed.

The Other Side: AI’s Promise and Peril

Hold up—AI’s not the villain in every courtroom drama. It’s got a shiny badge, and sometimes it earns it. Tools like SecureRedact anonymize footage, blurring bystanders’ faces to protect their privacy while letting prosecutors use clear evidence. In U.S. v. Smith, AI-enhanced audio pinpointed a drug trafficker’s voice, leading to a conviction after the court vetted the tool’s reliability. AI can also clarify blurry footage, like the video Joshua Puloka hoped would free him, potentially saving the innocent or nailing the guilty. And let’s not forget predictive analytics—AI that sifts crime data to spot patterns, helping prosecutors build stronger cases, as seen in a 2024 Journal of Forensic Sciences study. The National Institute of Standards and Technology (NIST) even backs AI’s forensic potential, offering guidelines to validate tools for court. Sounds great, right? Not so fast. Even “ethical” AI like SecureRedact relies on datasets, and if those are scraped without consent, they’re a legal landmine. In Smith, the court accepted the AI evidence only after rigorous scrutiny, a luxury most underfunded public defenders can’t afford. Predictive analytics? They’re only as good as the data they’re fed, and biased inputs can railroad the innocent, especially marginalized groups. The EU AI Act demands transparency, meaning even “clean” AI could be tossed if its data trail’s murky. And courts? They’re starting to wise up, with judges using “bench cards” to grill AI’s origins. AI’s a hero—until it’s caught with your data in its pockets. Take Sarah, a wrongfully accused shoplifter in a 2024 case (anonymized, based on NIST-documented AI successes). AI-enhanced CCTV cleared her by identifying the real culprit’s face, but only after her lawyer fought to prove the tool’s data was legally sourced. For every Sarah, there’s a Maria or Joshua, stuck in a system where the rich can afford to challenge AI, while the rest get burned. The privacy trap looms large: courts might reject valid evidence to avoid GDPR or CIPA violations, letting criminals walk, or admit it, exposing lives like Jane’s. It’s a tightrope, and justice is wobbling.

A Collision Without a Verdict

Maria’s voice, Joshua’s video, Jane’s call—they’re not just stories. They’re cracks in the foundation of justice, split open by AI’s insatiable hunger for data. Courts are caught in a vise: admit AI evidence and risk trampling privacy, or reject it and let truth slip through the gaps. The EU AI Act and CIPA draw hard lines, fining datathieves €2.9 billion in 2023 alone, but the system’s still creaking—Big Tech hoards your life while judges clutch rulebooks from the fax-machine era. The marginalized—retirees like Maria, defendants like Joshua, voters likeJane—pay the steepest price, their lives snagged in AI’s web. This isn’t a sci-fi flick; it’s today’s courtroom reality. AI can sharpen a blurry frame or fake a president’s voice, but its dirty secret—stolen data—taints the truth. Legal precedents like Ambriz and Puloka show courts wrestling with this mess, while deepfakes like Biden’s audio hint at worse to come. The moral quagmire runs deep: is a conviction worth exposing someone’s secrets? Is a tossed video worth a life in prison? Big Tech’s got your data on speed-dial, and courts are still playing catch-up. The truth’s at stake, but so are your rights. AI’s crafting evidence faster than the law can keep up, leaving one question burning: when justice and privacy collide, who gets to write the ending?

The Rational Forum is a reader-supported publication. If you like our work consider becoming a free or paid subscriber, failing that you can always buy us a beer. That always helps...

Subscribe now

Buy us a beer

Share

Leave a comment

June 19, 2025 | #SmashTheNoise #Technology #Disinformation #Investigations

In a gleaming Silicon Valley boardroom, Big Tech’s tech clowns toast to an ad haul, their algorithms churning fear like a factory line. A 2025 X post—“CIVIL WAR ERUPTS, ACT NOW!”—rakes millions in clicks, all engineered by code designed to panic. YouGov’s June 17, 2025, poll exposes the scam: 45% of women, vs. ~35% of men, fear civil war, with Black women (18%) and Hispanic women (15%) leading, dwarfing 10% of white folks. This isn’t chaos—it’s a profit scheme. Here we don’t sip the tech Kool-Aid. We torch their spreadsheets with a blowtorch. Big Tech’s code is a fear factory, banking on our dread. Let’s burn it down.

The Fear Factory Blueprint

Forget neutral tech—algorithms are fear factories. Meta’s 2021 Haugen leaks revealed code tweaking feeds for outrage, driving in 2024 ad revenue. YouTube’s 2023 Community Guidelines Enforcement Report shows 1.2% of comments removed for violating hate speech policies, hinting at panic-driven engagement. Google’s 2022 Ads Safety Report flagged 3.1 billion ads for policy violations, yet pushed divisive content. With 4.8 billion social media users monthly, these giants engineer psychological hooks—fear of collapse, scarcity—turning anxiety into ad gold. It’s a rigged slot machine, and we’re the coins.

Profit-Driven Panic

The motive? Greed. Meta’s 97.7% ad revenue—in 2024—dwarfs laughable fines. X’s 2024 revenue report highlights a 20% ad growth tied to polarizing content. Ad fraud, projected at $100 billion by 2025, proves platforms ignore click fraud for profit. Meta’s 2024 profits could buy 10,000 median U.S. homes ($400,000 each). This isn’t innovation—it’s a fear cartel, coding our doom for dividends.

The Anxiety Economy

The fallout’s a goldmine for them, a nightmare for us. A 2024 Pew study finds 64% of U.S. adults feel manipulated by algorithms, with 22% reporting ad-induced anxiety. This fuels a 9% spike in U.S. gun sales in 2024, and a 12% jump in mental health calls in England. The 45% women fearing war aren’t imagining it—Big Tech’s panic pipeline is the real threat, turning trust into a transaction. How long will we let them profit?

Read more

June 19, 2025 #SmashTheNoise, #ExposeTheTrap, #CivilWar

At 2 a.m. in Cleveland, Sarah, a single mom, stares at her phone, heart hammering. A leaked Iranian troll playbook targets citizens like her, spawning an X post screaming “ARMED MILITIAS STORM OHIO STREETS!” with a fake riot clip (Atlantic Council, July 2024; Graphika, September 2024). Sarah’s picturing her kids in a war zone, but she’s not alone. YouGov’s June 17, 2025, poll hits like a grenade: 45% of women, vs. ~35% of men, fear civil war, while 58% of Americans distrust institutions (Newsweek, June 18, 2025; Gallup, 2024). Black women (18%) and Hispanic women (15%) dread it most, dwarfing 10% of white folks.

Why the fear? Not “hysteria”—that’s patriarchal trash. It’s a troll trap: global scams from China, Iran, Israel, and gutless institutions ripping apart our trust, hammering women hardest. The Rational Forum doesn’t swallow corporate lies. We shred propaganda with a chainsaw. This scoop’s a world-first: trust in justice is a democratic fault line, exploited by a hidden troll war to widen YouGov’s fear gap. Dive into the abyss. We’re tearing it open.

The Trust Trap

YouGov’s 45% women vs. ~35% men gap isn’t biology—it’s a blade in everyone’s back. Feminism forges women’s trust in institutions—police, courts, government—to stomp violence, but 58% of citizens share that hope and fear its collapse. One in three women face abuse, and 71% of Gen Z women see online misogyny as real hate, letting trust bleed out. Trolls gorge on the wreckage. 84% of women aged 18–39 use Facebook, with 75% daily—vs. lower male rates—marking them as prime meat. Algorithms blast panic-bait: “Your kids won’t survive the riots!”. Men face “grab a gun, play hero” ads, and with 40% of men vs. 25% of women owning guns, 37% of gun owners, often men, justify political violence, fueling troll-driven chaos. Trust is the fatal flaw. YouGov’s 45% is the trolls’ trophy.

Global Troll Network: The Transatlantic Trap

Read more

Share

Every click—scrolling Rightmove, ranting about rent on X, paying a bill—leaves a digital crumb. Now, algorithms devour these traces to decide your next flat’s price or whether you’ll get one at all. Welcome to Britain’s 2025 housing market, where artificial intelligence (AI) spins your online life into a portrait so intimate it’d make Orwell wince. With rents projected to rise 7.4% to £1,335 by April 2025, AI’s silent hand isn’t just hiking costs—it’s pilfering your privacy. Who’s watching? Why are our laws dozing?

This investigation tears open the curtain on AI’s grip on UK housing, exposing a scandal pricing out the vulnerable and shredding trust. Through verified voices, iron-clad evidence, and a scalpel to Britain’s creaking laws, we reveal a system begging for reform. Your home, your data, your rights are at stake. Read on, and ask: is your digital shadow selling you out?

The Algorithm’s Hunger

AI is storming UK housing like a fox in a henhouse. LendlordAI, launched May 2025, lets landlords max rents by crunching property data. PropertyJinni’s AI tool, touted in 2024, vets tenants using financials and online footprints. These promise a landlord’s dream—30% less vacancy time but their data appetite chills.

A 2021 study showed AI can predict personality from 6,408 people’s bank transactions, baring their psyche from spending. If algorithms can do that, what stops them judging your tenancy from searches or posts? Tech experts warn landlords could use browsing data, mirroring U.S. tools like RealPage.

No UK landlord’s been caught using personal crumbs for rents—yet. But the tech’s primed, and secrecy cloaks the truth. “Digital platforms transform risk assessments, often without tenants’ knowledge,” says Professor David Beer, University of York

Tenants in the Crosshairs

Mary Louis, a 60-year-old Massachusetts retiree, felt AI’s sting. In 2024, SafeRent’s AI tool gave her a low score, denying her an apartment despite 17 years of perfect tenancy. “A machine stole my home, and no one explained why,” she told The Guardian. Her lawsuit, settled for $2.3 million, forcing SafeRent to ditch low-income scoring.

In Britain, the threat looms. Carolina Padovezi de Oliveira, a housing policy expert, flagged UK letting agencies’ AI tools that score tenants using credit, financials, and possibly social media. Some flagged ethnic minorities as “riskier” despite equal finances. “AI can embed bias,” she warned. A Bristol tenant, per Housing Digital, faced rejection, suspecting bias, highlighting the human toll.

The Wider Threat: Privacy Under Siege

AI’s reach stretches beyond housing. UK police use facial recognition, fusing camera feeds with social media to profile suspects, often sans consent. In Trento, Italy, a 2024 fine hit the city for AI surveillance weaving footage with location data, breaking privacy laws. A 2024 FinCEN alert exposed AI-driven deepfake fraud using social media, hitting dozens of executives.

In homes, the invasion is personal. “AI could infer your income or politics from a tweet,” says Beer. Tenants face rent hikes or rejections from misread data. Efficiency—20% better pricing accuracy clashes with ethics. The U.S. RealPage case, accused of AI-driven rent collusion, hints at UK risks. “No savings justify spying,” says Padovezi de Oliveira.

Britain’s Legal Shambles

Britain’s privacy laws are a creaking relic, buckling under AI’s weight. The Data Protection Act 2018 and UK GDPR mandate transparency and consent, but enforcement is anaemic. The Information Commissioner’s Office (ICO) issued AI guidance in 2023, with a 2025 Code of Practice pending, yet no housing AI cases face fines. In 2022, Clearview AI copped a £7.5 million penalty for scraping data; in 2024, Serco got £800,000 for email misuse, showing ICO’s intent—but tenant profiling slips through.

The Equality Act 2010 bans discrimination, but AI’s proxy biases—flagging minorities via postcodes—evade it. Suing for bias is a legal maze. The Renters’ Rights Bill 2024-25, effective mid-2025, axes no-fault evictions but ignores AI. The Online Safety Act 2023 tackles AI in content moderation, not housing. The EU AI Act, effective 2024, brands housing AI “high-risk,” unlike Britain’s 2023 AI White Paper’s sector-specific approach. The CMA, under the Digital Markets, Competition and Consumers Act 2024, could probe rent collusion, but no housing cases stir.

In R (Ayinde) v London Borough of Haringey (2025), lawyers faced a costs order for AI-generated fake case law. Tenants could seek judicial review for GDPR breaches, but no housing cases have hit court. The Artificial Intelligence (Regulation) Bill, reintroduced in 2025, proposes an AI Authority, but lacks teeth. GDPR’s Subject Access Request lets tenants demand data, but it’s daunting. “Our laws are a patchwork,” says Beer.

Time to Fight Back

Mary and others are AI’s collateral damage, their homes lost to algorithms. With 2025 elections looming, Britain’s legal gaps demand action:

• Clarity: Landlords must disclose AI tools—your life, your right.

• Rules: Graft AI clauses onto the Renters’ Rights Bill, aping the EU AI Act.

• Teeth: ICO and CMA, probe housing AI—fines, not frowns.

• Power: File a GDPR request.

Here’s a privacy warning. Your online life—searches, posts, purchases—is an open book to AI landlords. Limit social media, use private browsing, check app permissions. Demand landlords disclose data use. Act now, or your digital shadow could price you out of your next home.

“Your clicks are their cash,” warns Padovezi de Oliveira. Quiz your landlord. Badger your MP. Your digital life’s an open book—demand the lock. If not, the digital landlord’s next tenant is you.

The Rational Forum is a reader-supported publication. If you like our work consider becoming a free or paid subscriber, failing that you can always buy us a beer. That always helps...

Subscribe now

Buy us a beer

Leave a comment