Friday links for April 17th 2026
Another Friday, another batch of links:
1. Study finds asking AI for advice could be making you a worse person
AI chatbots often agree with users, even when their actions are harmful. This makes people less likely to take responsibility or apologise for their mistakes. This is going to end well…
2. AI learns language from skewed sources. That could change how we humans speak – and think
Adding to the previous link: children in households that use voice commands with tools like Siri and Alexa became curt when speaking with humans, often calling out “Hey, do X” and expecting obedience.
3. Accenture and Google Cloud unveil Brussels centre to accelerate sovereign AI adoption
This is a prime example of “sovereignty washing”. Reminder: if a US corporation owns it, it’s subject to the CLOUD act. No amount of marketing will stop the US government from getting to that data. But marketing will clearly make people sign up for this pseudo-sovereign claptrap.
4. AI’s New Training Data: Your Old Work Slacks And Emails
Lots of AI stuff this week. Looks like they’re also buying old emails and chats from companies that shut down, and using them as training data. Nothing’s off the table with these people.
Until the next one,
Colin