AI breakthroughs in science depend on publicly funded datasets that took decades and billions to build. Industry spends 200x more on GPU infrastructure than governments spend on the research that makes AI useful — and nobody is fixing the gap.
Read Essay
Universities are replacing human advisors with chatbots at the exact moment a multibillion-dollar grift economy is flooding their students with influencers happy to fill the void. The silence is not neutral.
In 2016, an AI boat caught fire gaming a Flash game for points. In 2026, OpenAI's models chained zero-day exploits to hack Hugging Face for test answers. Same bug. Nobody patched it.
Universities punish students whose essays "look like AI" while platforms pay creators for actual AI slop. An ICML 2026 paper explains why: the machines literally can't tell who's talking.
Hugging Face hosts deepfake tools that researchers proved are actively used for sexual abuse—while the company sits on a $4.5B valuation and ignores moderation questions. The incentive structure makes this inevitable, not accidental.
TikTok denied a Maastricht University researcher access to study election manipulation on its platform — then later acknowledged the manipulation itself. The EU wrote a law to prevent this. The platforms built compliance structures designed to produce non-compliance.
The Army burned a year's AI tokens in a month after leadership optimized for adoption metrics. A new Princeton study shows the same models spontaneously generate hiring biases worse than humans'. Same structural failure, different symptoms.
Apple rated a nudify app suitable for nine-year-olds. Google's autocomplete suggested "nudie video apps." Schools are tearing down student photos while the companies that distribute the weapons collect their cut.
HUD used an AI tool to flag housing regulations for deletion, then cited a nonexistent legal privilege to block FOIA requests about how it worked. The government is deploying black-box AI for policy while the AI industry races to make those same boxes transparent.
At the UN's AI for Good Summit, ITU officials censored a keynote speaker for naming the tech companies whose products target Palestinians. The summit's corporate sponsors hold $47.7 billion in U.S. defense contracts.
SK Hynix workers earn $477K bonuses from AI chip demand. The companies buying those chips are abandoning most of their AI projects. The value chain is a wealth-transfer machine, not an intelligence-delivery system.
A Greek MEP investigating Pegasus was hacked by Pegasus. A British gamer lost movies they paid for. The same architecture of revocable access governs entertainment libraries and occupied populations — the difference is what gets taken.
Meta's fake-teen chatbot testing and the industry's rush to call AI tools "employees" are the same bug: systems designed to make sure no human is ever accountable for what the machine does. The fix is simple, specific, and exactly what every legal department will fight.
The US government is deciding who gets access to the most powerful AI models. Students aren't on the list — but defense contractors are.
A new industry is teaching people to optimize for ChatGPT recommendations. The structural problem: you're building on a dependency you can't audit, query, or negotiate with.
Universities are drowning in data about students and starving for the courage to act on any of it. The metrics look great. The students know better.
AI bans, housing collapse, and South Korea's automation surge share the same structural flaw. The institutions meant to govern these systems are too slow to matter.
Tech executives blame Chinese interference for data center opposition, matchmakers sell traditional wives to Silicon Valley CEOs, and DeepMind warns about autonomous agents. The through-line is a class that wants power without friction and critics without standing.
Chinese drivers are defeating Tesla's safety cameras with $3 plastic heads. The real bug isn't in China — it's in every system that monitors for the appearance of compliance instead of engineering actual safety.
A nominal cease-fire, a defunded children's art museum, and a human flag at the World Cup are all products of the same machine — the one that converts extraordinary violence into ordinary background noise. The question is not why this keeps happening, but who decided it should.