There’s a new Android phone hack that lets anyone with your phone send messages as you from the lock screen — without tapping in your PIN number. Why? Gemini! You grab someone’s phone, press the power button — now the Gemini button — and try to send a message: [Register] Selectin … | Continue reading
Rodrigo Ghedin translates “Today’s stupid prompt injection: just write like the AI” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
Today we have another interview. I’ve got Zencastr working, so I want to do these regularly. Please suggest interviewees! This is once again with our good friend Kim Crawley, promoting her forthcoming book Recovering After GenAI. It’s on the podcast and the YouTube. [LibSyn; YouT … | Continue reading
When you put text into a large language model, it responds with some words. You make an LLM into a chatbot by feeding it the whole conversation so far with your prompt on the end, and it responds to that whole text with whatever it thinks comes next. A reasoning chatbot uses a mo … | Continue reading
I went on the Times Radio Tech Report again today: “AI Bubble: At some point the subsidies run out”. This time I got the better camera working! [YouTube] I’m interviewed on the Fault Tolerant substack: “Smoke & Mirrors & LLMs.” [Substack] I’m doing a web seminar with our good fri … | Continue reading
Meta tried buying their AI-powered creep glasses a public image with celebrity sponsorships — but the celebrities’ fans hate them. So do the other celebrities. Engadget spoke to a pile of photographers who’d been trying out the glasses. They like the cameras and the convenience — … | Continue reading
The UK Government will speed up justice — with AI! UK police will take evidence in criminal cases and … run it through the chatbot. [Gov.UK] Under current rules, police have to analyse each piece of evidence themselves before disclosing it to the defence. The Home Office will: al … | Continue reading
Today we have a report by Antonia Juelich at the Cambridge Programme on AI Science and Policy, CASP: “‘God has helped us, and so will AI’: How the Terrorist Group Boko Haram Uses Frontier AI”. Oh no! The bad guys have AI! [CASP, PDF, archive] CASP’s organisational mission is to p … | Continue reading
Rodrigo Ghedin translates “Ecosia’s odious greenwashing — now with AI” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
NeurIPS is an annual machine learning conference. Here in the AI bubble, NeurIPS is huge and flooded with chatbot cash. [Atlantic, archive] The AI vendors at NeurIPS want to talk about imaginary problems, like the super-AI destroying humanity. The researchers want to get on with … | Continue reading
Meta’s AI glasses have all sorts of alleged use cases. The real use case is the market of creeps and weirdos who film people out and about without their consent. And Meta knows for a fact this is the market: [Meta] Beginning with our second generation of glasses, the camera is au … | Continue reading
Isaac Pound from Times Radio’s Tech Report interviewed me this afternoon talking about AI and the job market — and it’s up already! 32 minutes of me talking quite fast. This came out super well. [YouTube] The camera is an old Logitech web cam — the one I plan to do interviews on … | Continue reading
The AI bubble is losing steam. Venture capital is desperate for a new lottery ticket. What’s the next bubble after AI falls over? Will it be modular nuclear? Will it be humanoid robots? The hot favourite is quantum computing! Quantum will break all the codes! The obsolete ones. I … | Continue reading
In 2017, a team at Google Translate came up with a new machine learning architecture: the transformer! Transformers are just ridiculously effective for machine learning. Google Translate itself switched to transformers in 2020, and straight away it got a lot better. You know one … | Continue reading
The UK announced its AI Opportunities Action Plan in January 2025 — a signature programme for the new and otherwise ideas-free Labour government. It would include a huge OpenAI data centre or two, part of OpenAI’s Stargate plan! £20 billion of new UK data infrastructure! In Febru … | Continue reading
Rodrigo Ghedin translates “AI leaks your company’s code secrets faster than ever” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
Ecosia is a search engine that talks up its green credentials. It repackages Microsoft Bing and puts its own ads on. All the money comes from those ads. Ecosia’s big pitch is those ads pay for planting trees. That’s fine! Unfortunately, Ecosia has something extra. It’s an AI. Wha … | Continue reading
Frontiers is an academic publisher that keeps showing up on lists of predatory open-access publishers, whose business model certainly appears to be to accept any old rubbish and charge a swingeing fee to the author. Frontiers does try to keep a slight scientific gloss by recruiti … | Continue reading
When you write a computer program, you will often want it to access private things, like a database or a service you’re paying for. A human, you could ask for a password. But a program needs an access secret so a human doesn’t have to press a button each time. There are ways to d … | Continue reading
Last year, director Luca Guadagnino signed up for “Artificial,” a movie about Sam Altman and that weekend in late 2023 when the OpenAI board kicked Sam out for being a serial liar. [Deadline, 2025] It was going great! This was going to be a good movie! Then Amazon abruptly pulled … | Continue reading
Anthropic’s been bigging up its new Mythos and Fable models. Mythos can hack anything, you know! So can Fable! They’re just too powerful and dangerous! So our KILLER ROBOTS can definitely answer your office email. Earlier this month, one large organisation took Anthropic’s warnin … | Continue reading
Rodrigo Ghedin translates “Office workers are spending way too much on AI too” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
Masayoshi Son is the founder and CEO of SoftBank, the giant Japanese private equity fund that’s the main backer and cheerleader of OpenAI. SoftBank was also the main backer of WeWork, which was almost as stupid as OpenAI. WeWork’s 2019 stock market offering failed spectacularly a … | Continue reading
Every month on the podcast and YouTube, I do a patron name read. This month’s is Tuesday 30 June. If you’re on the Patreon, go here and comment with your desired name to read out. If you’re a YouTube member, email me to be listed. If you’re not on the Patreon, click here! Deadlin … | Continue reading
The AI push runs on setting venture capital money on fire and charging chatbot users way less than the bot costs to serve. The idea is that the customers will get hooked on the business value. Then the AI vendors can gouge them. And the vendors can try not to haemorrhage quite so … | Continue reading
Pivot to AI relies on support from you to keep bringing the quality AI bubble foolishness, five days a week. It’s time to rattle the tin again. Keep Sam Altman crying eight delicious and bitter tears for every dollar you send to Pivot to AI: The Patreon is here. Sign up for $5 a … | Continue reading
Meta announced a fabulous new programme in April — the Model Capability Initiative! [BBC] a new tool will run on Meta’s computers and internal apps, logging their activity to be used as training data for AI technology. This collected every keystroke, mouse click, and screen image … | Continue reading
Last July, publishers filed competition complaints about Google’s AI Overview in the UK and Europe. The overview directly substitutes for the publishers’ own sites with a worse version. And you can’t opt out of the AI Overview without blocking Google Search. Until now! The UK Com … | Continue reading
The Pope put out his encyclical about AI in May. But Catholics weren’t first to claim a religious exemption from AI – it was a Unitarian Universalist. [Business Insider] Erin Maus is a software engineer in North Carolina. She works for an large “tech-entertainment” company that s … | Continue reading
Rodrigo Ghedin translates “We found the missing AI apps! And nobody downloads them” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
Midjourney is an AI image generator and copyright violation machine. What’s the weirdest and dumbest business pivot an image generator company could possibly make? Yesterday, Midjourney CEO David Holz announced Midjourney Medical! No, they didn’t call it “Medjourney.” They’re goi … | Continue reading
Unreal Engine is a video game engine produced by Epic Games. It’s what Fortnite is written on top of. Epic CEO Tim Sweeney is convinced the slop machine is the future. In November, he was claiming the Steam store — Epic’s direct competitor — should stop tagging games using AI, be … | Continue reading
Chatbots are a security hole the moment you let them do real work — because you can always prompt-inject a chatbot. And if Microsoft does this stupid thing, you can break in and get confidential company information. This one comes from Varonis, who found the “reprompt” attack in … | Continue reading
Last September, programmer Mike Judge asked: if AI’s so good for coding, where are all the little AI apps? Where are the widgets someone just vibed? Where’s all the AI shovelware? He worked out the numbers — and there had been no uptick in apps going to the Apple or Google stores … | Continue reading
Google’s AI overviews are accurate 91% of the time! Which means 9% of the time, the AI is just making stuff up. [NYT, archive] That’s not good enough when the AI hallucinates that someone’s a crook. Two publishing houses in Munich sued Google for defamation when the AI overview m … | Continue reading
Rodrigo Ghedin translates “rsync goes AI slop, breaks your backups” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
We are moving house from London to Glasgow over 10–11 June! Current plan is no Pivot to AI for 4–12 June. I should be back Monday 15 June. You’re on your own until then. Get out there and be rude to a computer today. | Continue reading
rsync is a program for copying entire file folders from one computer to another, doing incremental updates if there’s only a few changes. You can keep a continuous backup copy. Rsync is super reliable for this job. Until recently. On 28 May, Jeremiah Fieldhaven posted about how h … | Continue reading
In March, Meta started using chatbots for customer service — the Meta AI Support Assistant. [Meta] With the Support Assistant, Meta has a new workflow for Instagram account recovery: Work out where the owner of an Instagram you want to get into lives in. Log in to a VPN with an e … | Continue reading
You can prompt-inject ChatGPT by asking it to summarise a web page for you. Andi Ahmeti from Permiso got ChatGPT to display his chosen phishing link or QR code in ChatGPT’s own output to the user — so an evil link looks like it’s a system message from ChatGPT itself. [Permiso] Ho … | Continue reading
Rodrigo Ghedin translates “The SpaceX IPO works like a crypto fraud, but with AI” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading
Jqwik is a test engine for Java programs. Jqwik does not welcome AI coding: [GitHub, archive] the copyright consequences of training an LLM with mostly public code repositories have not been clarified. Moreover, hyper-scaled GenAI is a fundamentally unethical technology that shou … | Continue reading
We’ve seen this film before. Before our pivot to AI, we wrote about cryptocurrency fraud. A crypto initial coin offering would start with a white paper made of impossible lies. Nobody cared because the entire value proposition was “number go up.” The crypto coin would launch, the … | Continue reading
McClatchy Media runs newspapers around the US. But the management have fallen to the AI. Their exciting new plan is called “content scaling”. From Eric Nelson, vice-president of local news: [Wrap, archive] Nelson was promoting the company’s new “content scaling agent,” an AI summ … | Continue reading
Pope Leo XIV presented his first encyclical letter to the Vatican on Monday. Titled “Magnifica Humanitas,” or Magnificent Humanity, the 43,000 word missive specifically takes aim at AI — and the AI bros. [Vatican] It’s a book-length writeup on the AI bubble and why it’s terrible. … | Continue reading
WEB PAGES WERE NOT MEANT TO BE FOUND YEARS OF SEARCHING yet NO REAL-WORLD USE FOUND for PAGES of SEO-OPTIMISED AI SLOP Wanted to find things anyway for a laugh? We had a tool for that: it was called WIKIPEDIA “RANDOM ARTICLE” “Yes please give me ONE HUNDRED AND TEN MILLION of my … | Continue reading
We have the flat in Glasgow! My life is money-with-wings 💸 emojis. The BIG MOVE is 10–11 June. Current plan is no Pivot to AI for 8–12 June. I should be back Monday 15 June. Cross fingers! Thanks for all the $5/month Patreon signups and Ko-Fi donations — they’ve been vita … | Continue reading
Rodrigo Ghedin translates “GitHub Copilot AI token charges to go up 10×–100×” into Portuguese for Manual do Usuário. [Manual do Usuário] | Continue reading