AI Slop, Secret Hacks and Student Freebies: AI Haven’t a Clue’s weekly round-up

Ransomware gangs threatening to feed stolen art into training data. Google throwing freebies at students and judges sitting on billion-dollar copyright payouts.

The Alan Turing Institute: From Beacon to Battlefield

We’ve been tracking the drama at the UK’s flagship AI research hub for weeks. Now the CEO has resigned after whistleblower complaints and a 93-staff letter of no confidence. The government is leaning hard on the institute to focus on “sovereign AI capabilities” and defence. Once upon a time it was about health, environment and responsible AI. Now it’s pivoting to military priorities.

Ransomware with an AI Twist

Lunalock is a ransomware gang that hacked the art marketplace “Artists and Clients” and is demanding $50,000 or it will leak stolen works - and feed them into AI training datasets. As James points out, this is worse than just leaking art. Once it’s in a model, it’s there forever. Artists are fighting back with tools like Glaze and Nightshade which subtly poison images so AI can’t train on them. Your eye sees a man on a horse, the model sees a dolphin in a pond. It’s watermarking as sabotage and it might become essential.

The AI Podcast Factory Nobody Asked For

A startup called Inception Point AI claims it can produce 3,000 AI-hosted podcast episodes a week at a cost of one dollar each. They say the “age of AI slop is over.” George calls it exactly what it is: AI slop. No quality control. No passion. No human spark. They even want to stop flagging which shows are AI-hosted. James sees a niche use case for hyper-local shows. George hopes it crashes and burns.

Anthropic’s $1.5 Billion landmark case

Anthropic, maker of the Claude LLM, agreed to pay authors $1.5 billion in what would be the biggest copyright recovery in history for training on pirated books. But the judge has postponed approval, creating confusion. If it sticks, it could open the floodgates for claims against every big AI firm. Judge Alsup knows the stakes. This is one to watch.

Secret AI at Work

A new US survey shows AI job anxiety rising at all levels, especially among executives. People are secretly using AI to look good at work and not disclosing it. The top reason? They like having a secret advantage. James says transparency and clear company policies are the way forward. The hypocrisy of insisting artists label AI work while hiding your own ChatGPT use wasn’t lost on them either.

Good AI News (Yes, It Exists)

Apple’s new AirPods promise real-time translation. At the same time, African Next Voices has created the largest dataset of African languages to unlock AI for underserved communities. The Bill Gates Foundation helped fund it. It’s a glimpse of AI breaking barriers rather than making them.

Trump’s AI Power Play

Trump invited Bill Gates, Sam Altman, Mark Zuckerberg and Tim Cook to a dinner about America’s AI future. Elon Musk was notably absent. Trump is pivoting from industry and factories to big tech, as his ticket to global leadership.

Google’s Freebie for Students

Google is offering UK students a free year of Gemini Pro, including NotebookLM, video tools and unlimited chats.

Vegas Goes AI Wizard of Oz

The Las Vegas Sphere has reimagined The Wizard of Oz with new AI visuals. Flying houses. 3D tornadoes. Glitter everywhere. Some say it’s magical, others call it a violation of the original creators’ intent.

Meet Abdu Tahir: Impact First, AI Second

Abdu runs Drift Time, an agency helping people and organisations working on sustainability and social impact. He’s experimenting with AI but only where it actually helps. His takeaways:

• Use AI intentionally. Don’t waste energy on trivial outputs.

• It’s a tool, not a replacement. Every creative wave has had its “Macintosh moment” and you adapt.

• AI raises the floor for small players but risks eroding craft and resilience. Without learning the hard way, people may lose grit.

• Prompt engineering is essential. Learn to drive the tool before it drives you.

Wrapping Up

If you’ve got an AI question for James, email aihavenaclue@gmail.com… Yes, mugs are still being awarded for the best listener questions.