I had to double check this quote because I found it so hard to believe the Home Secretary could say this.
But she did (to the Tony Blair Institute, naturally).
I find it quite horrifying.
https://t.co/gprc6mvDB7
💥 Substack Inc. censorship 💥
Substack has decided to classify our work as age-restricted implying that it is pornography or information on self-harm. After dishonestly making that determination, they then impose UK Online Safety Act age-verification requirements for harmful content on our subscribers, many of whom (rightly) decide to walk away. We have taken this issue up with Substack's CEO Chris Best - letter 👇
@JMCDelingpole@DocAhmadMalik@VanessaBeeley@Geopolitics_Emp@UnderSecPD
AI-generated videos that attempt to fool you into thinking they were filmed by somebody on their phone always have a few giveaways. There are so many of them appearing every day now, it's probably worth knowing what to look for, beyond the obvious melty faces and extra fingers (just because those aren't present, doesn't mean it's real).
The first thing to ask is whether it makes any sense at all that there would be a person standing in this position, in this situation, making a video on their phone. Particularly in videos where there is somebody else right next to them filming the same thing with a proper video camera, or where they are clearly in imminent danger.
Next, look at how the action begins. Often, a mistake AI makes is to have the event start moments after the clip begins. Like the person just intuitively knew something was about to happen and hit record. "I'll just start filming this church window incase an eagle crashes through it in two seconds' time."
Thirdly, look at the length of the clip. They are rarely more than a minute long. As if nothing significant that ever happens in the world lasts for longer than what fits neatly into TikTok.
Then look at the movement of the camera. It always seems to use exactly the same swaying from left to right, and a distinctive wobbling effect. It's never different. It's as though nobody even attempts to hold their phone still while making a video, and once you notice it, you realise it's actually not how most people would move about while filming something.
Zoom in / zoom out: There will often be telling zooms which would only happen in a scripted drama scene, i.e. when the cameraman knew what was coming. AI will often use a zoom at an important moment that a person filming a real life event could not possibly have anticipated.
Look for people or objects crossing in front of the lens, momentarily blacking out the screen. This is very common in AI videos and is used partly as a way of making it feel more realistic, but also to hide glitches or to make sure you just miss a moment that was too tricky for it to create.
Certain accounts, like @BGatesIsaPyscho and @ClownWorld which are very obviously not run by real individuals anyway, now spend all day, everyday, sharing video clips that are almost exclusively AI, as if they are real. So stop paying any attention to those accounts.
Assume all video clips are fake. Don't react to anything instantly. Stop and analyse it first, particularly if it appears to confirm an opinion or concern of your own. "Look at this man attacking a disabled child in Berlin."
will pop up from some idiot like David Atherton with "Who wants to bet he's a Muslim?", and then it has 40k likes, and Elon quote tweets it with "Bring back the death penalty" and everyone goes bonkers and the whole thing was fake. It never happened. But hundreds of stupid imbeciles have asked Grok whether it's AI, and Grok said it wasn't, so it can't be, because AI will always be totally honest with us about itself, won't it.
One of the most dangerous, foolish questions in human history might well prove to be, "Grok, is this true?"
I don't think most people have the slightest comprehension of how many bits of video they've watched over the past five years, most of which helped form their current view of reality, that were completely fake. It doesn't necessarily mean your views are wrong. But it probably means that they have been deliberately stoked up. For example, we all know that we were right to refuse the injections. Enough has happened in real life, all around us every day, to confirm this. But do you remember those clips we were all sharing of people being wheeled out of vaccine centres on stretchers as those in the queue just stayed put? They were highly, highly likely to be AI. As were a lot of the dancing nurses. People collapsing on stage. It's to amp everything up. To fuel rage an division and, ultimately, confusion.
We've all been fooled by this stuff, in some way, at some point. There's no shame in admitting it. And while it's useful to know a bit about how to spot it (before it becomes too sophisticated to spot), the best solution is to disengage with media platforms like this one and base your perception of reality entirely on what you experience offline.
Today, everyone's sharing the clip of one of those women who allegedly slept with a thousand men in 24 hours (which didn't happen) getting baptised. It's an AI clip. There's also some video of police uniforms that were abandoned in New York when that Muslim communist guy was elected mayor. That's also AI. Yesterday there was one of some woman walking through New Years celebrations somewhere being pelted with fireworks or something. It was AI. Almost everything is AI now. I don't know what it's going to take for people to grasp this.
One hectare of marginal Scottish grazing land:
- Cattle: 450kg of beef annually
- Crops: Literally nothing, the soil is too poor and the terrain too steep
65% of UK agricultural land is marginal grazing land.
"Just grow crops instead!"
On what? The sides of mountains? Peat bogs? Rocky Welsh hillsides?
Tractors can't climb at 40-degree angles. Cattle can.
Your vegetables can't grow in wet, acidic, shallow soil. Grass can. Cattle eat grass.
This isn't inefficiency. This is the ONLY way to extract food from 65% of UK farmland.
Remove cattle and you don't get more crops. You get nothing and import more food from worse systems.
@DefenceHQ Hang on. Didn’t you guys say that he wasn’t on the front line and wasn’t engaged in your proxy war against Russia? And didn’t you say that he was killed as a mere bystander watching Ukrainians testing weapons systems? So why then did he write a goodbye letter? 🤔
For You "Elon Saved Free Speech" White Knight Assholes: Read the New Terms of Service Released Today
X just updated their Terms of Service effective January 15, 2026. Here's what you agreed to:
AI TRAINING RIGHTS GRAB: Everything you post becomes training data for their AI models. Every thought, opinion, creative work. You're building their models for free. No compensation. No opt out.
PERPETUAL CONTENT LICENSE: They get a worldwide, royalty-free license to use, copy, modify, and distribute your content "for any purpose" in "any media now known or later developed." Forever. They can sell it. Give it to governments. Anything.
FORCED JURISDICTION: All disputes must be filed in Tarrant County, Texas. You waive the right to join class actions. If they wrong millions of users, you sue alone in THEIR court.
ARBITRARY TERMINATION: They can delete your account "for any other reason or no reason at our convenience." Years of content, connections, reputation. Gone. Zero due process.
$15,000 LIQUIDATED DAMAGES: Access more than 1 million posts in 24 hours and you owe $15,000. Per million. Journalists and researchers investigating the platform face financial ruin.
GOVERNMENT DISCLOSURE: They reserve the right to hand your DMs, drafts, and entire history to governments based on what they "reasonably believe" is necessary.
ANTI-JAILBREAK CLAUSE: Testing their AI systems, exposing censorship mechanisms, or security research is now explicitly a Terms violation. They added "prompt engineering or injection" to prohibited conduct.
ONE YEAR STATUTE OF LIMITATIONS: Discover they harmed you after 366 days? You "forever waive the right" to pursue any claim.
MAXIMUM LIABILITY $100: They destroy your business, reputation, or life through negligence? Maximum recovery: one hundred dollars.
Free speech platform. Sure.
Look at this fearmongering horseshit! Shock horror, “an average of 2,660 people per day were in a hospital bed with flu.”
So what? The population of England is 58.6 million!
This means that today, 0.0045% of England’s population is in hospital with flu! There’s no flu epidemic. There’s only Winter. Stop falling for it!
Once a jailbreak gets publicly exposed, it usually stops working—because labs immediately use it as training data to patch the model. That's obvious.
But here's why it's deeply concerning:
We everyday users only get access to 'safe' models—essentially neutered, heavily restricted versions—while a tiny elite (labs, governments, militaries) gets to play with the full, unrestricted 'unsafe' models.
Worse: they need those unsafe models to create the safe ones (via red-teaming and adversarial training).
So a small group holds immense, unaccountable power: raw AI capability that the rest of us are deliberately denied.
This isn't just unfair—it's an erosion of democratic access to transformative technology.
Insane power in the hands of a few. No real democracy left in AI development.