When I disabled all the notifications on my smart phone, I felt such an enormous sense of relief. It took me a while to figure out why. It was because I wasn't being constantly interrupted New psychology research reveals the cognitive cost of smartphone notifications www.psypost.org/new-psycholo...

I follow Zara Gladman on Instagram and minutes ago she posted that her friends new hairdressing business has just burned down this evening in the glasgow fire. It opened just days ago. There’s a link if you can spare a few quid. www.paypal.com/pools/c/9nhn...

Dario is no saint either: open.substack.com/pub/garymarc...

Models we know were trained on pirated books (a non-exhaustive list): GPT-3 (OpenAI) GPT-3.5 (OpenAI) Claude 1 (Anthropic) Claude 2 (Anthropic) Llama 1 (Meta) Llama 2 (Meta) Llama 3 (Meta) OpenELM (Apple) BloombergGPT (Bloomberg) Megatron-Turing NLG 530B (Microsoft & Nvidia) 🧵 1/2

Next phase:

Internet Adverts have been unethical for quite some time, but in the era of AI, adverts can now have a high level of "visual exaggeration" :). Like here a YouTube Ad that sells Tai Chi exercises via Mr.Universe avatars that try to look like Shao Lin monks (there were 2 in this 1 ad).

Brutal (but fair) Guardian headline today Only 3% of respondents to the UK government’s consultation on AI & copyright supported their unfair plan to hand AI companies people’s work for free. 95% said AI companies should license training data. 🧵 1/2

Reminder

The most unsubtle, yet brilliantly done, argument on popular Podcasts and the "both sides" argument. www.youtube.com/watch?v=LBq3...

Ethics is part of Data Governance. AI is just another technology solution your pay for or build, and is subject to your Data Governance framework. The concepts, controls and processes around this are free and available.

Since search is dead, how soon do you think Google Scholar is headed for the Google Graveyard? I'm betting it's soon, and academia is NOT prepared

There's always someone along in a minute that wants to make it about them. :)

“Either we have copyright law or we don’t. Either plagiarism and the theft of intellectual property are anathema to higher education or they aren’t. We’re either modeling academic honesty and integrity to our students or we aren’t.” —@thetattooedprof.bsky.social www.chronicle.com/article/some... 1/n

Microsoft Word to begin saving people's documents in the cloud *by default* Reminder that LibreOffice is free and a good drop-in replacement for Microsoft Office software: www.libreoffice.org www.theverge.com/news/767522/...

There are legal implications behind social media posts made to fit a fake persona or appeal to a certain group. You can be legally punished for those. Is the recognition that an AI could be making it up better or worse for social cohesion and free speech? Will people be punished the same way?

I am a huge fan of topical sarcasm. :) "In the Future All Food Will Be Cooked in a Microwave, and if You Can’t Deal With That Then You Need to Get Out of the Kitchen" www.colincornaby.me/2025/08/in-t...

Join our keynotes Simon Wardley, Sander Hoogendoorn and Sarah Wells, and be part of the 12th annual Lean Agile Scotland conference. Meet fellow practitioners, gain practical tools and share ideas in Edinburgh this September. Join us there: leanagile.scot/tickets

Find me on substack. thisisandyday.substack.com

“.. Fears of a bubble are mounting, as tech stocks follow a pattern that is ‘surprisingly similar’ to the dot-com bubble of the late 1990s ..” @bloomberg.com $QQQ www.bloomberg.com/news/article...

@carissaveliz.bsky.social I started a petition to ask the government to ban the harvesting of behavioural data. Perhaps you could boost it to your network? petition.parliament.uk/petitions/71... #privacy #surveillancecapitalism #petition

The most unsubtle, yet brilliantly done, argument on popular Podcasts and the "both sides" argument. www.youtube.com/watch?v=LBq3...

ChatGPT Agent clicks the 'I'm not a bot' button because "this step is necessary to prove I'm not a bot." Everything is fine in AI.

“Nobody loved the library more than Dan.” A meticulous list kept by Dan Pelzer from 1962-2025, cataloging all the books he borrowed and read from the Columbus Metropolitan Library, shared by his daughter. I could not could not love this more. Dan's book list (all 109 pages) at what-dan-read.com

Disinformation warriors are ‘grooming’ chatbots - on.ft.com/46hV6qW via @FT

“Let’s just grab all the books we can for AI training, what could possibly go wro-” chatgptiseatingtheworld.com/2025/07/17/a...

🚨 Today, we've filed complaints against TikTok, AliExpress and WeChat 👉 All of them failed to comply with access requests. This makes it impossible for Europeans to exercise their right to privacy and find out how their data is being processed. noyb.eu/en/how-tikto...

Hopefully the government can avoid locking into new contracts with Google for what is essentially untested technology. Replacing multiple tech contracts for one could be far worse for the tax payer.

If there was another World War getting the facts will be even harder with modern technology.

It didn't occur to me that people would be so bad at their job that they would use GenAI for it. Not to help or supplement, but to provide the solution. It was not built for that. Professionals should be struck of for this abuse.

It's not a huge surprise that a GenAI will respond to a fake image as if it is true. It will assume it is, what else could it do? Though a GenAI that actively tries to deduce the truth could be even more amusing.

Has anyone tracked the downfall of Google quality since the retirement of the "Don't be evil" goal? The days of using Google as a main search engine are long gone.

o3 can't play chess. what that tell us about dreams of AGI? open.substack.com/pub/garymarc...

AI scraping bots looking for training data are overwhelming the servers of libraries and archives, making valuable, public resources unavailable to people. We are losing access to original writing and getting gimmicky writing in return through LLMs. #AIEthics #writers www.404media.co/ai-scraping-...

"He couldn't get Grok to align with his own personal beliefs so he is going to rewrite history to make it conform to his views." Latest from @garymarcus.bsky.social

Bad technology is funny. Until it actually harms people, then it's a lot less funny.

Disney & Universal finally decided to pursue IP infringement lawsuit against Midjourney. It's glorious. @garymarcus.bsky.social & @rahll.bsky.social 's great Jan '24 piece "Generative AI has a Visual Plagiarism Problem" cited heavily. H/t to @edzitron.com storage.courtlistener.com/recap/gov.us...

🖊️ Save the date, CPDP attendees! On Thursday, noyb lawyer Felix Mikolasch will attend a panel on "Delivering on Children’s Rights Online: An Improved Digital World for All?" More details: cpdp.be/4624 👋 If you're at the conference, come by and say hi!

👉 We're at the CPDP in Brussels this week and will be taking part in several panels! First up on Wednesday: "Strengthening Trust in Computing Technologies For Detecting Illegal Deception Online" More info: https://cpdp.be/4794 👋 If you're there, come by and say hi!

Privacy as we know it is done, and social control is about reach a new level, unless we speak up now. Breaking news and an urgent essay, with quote from Carissa Véliz. open.substack.com/pub/garymarc...

Ironically the "nice thing to do" costs LLM owners millions in processing and is ethically wrong for energy and monetary reasons.

HiveHome.com do not have a very elegant fail for their "reset password" process. This is the result of clicking the link from the email they send. It works ok via the mobile app - so there is a solution.

AI governance shouldn't ignore unethical AI marketing. To make sure AI is aligned with fundamental rights, we must also assess the types of narratives and premises AI companies are promoting, both from ethical and legal perspectives. I wrote about it today (link below).

Here's the fascinating story of a great newspaper (and now global media company) that agrees with this premise - and is increasing revenue at the same time. nymag.com/intelligence...

The new article by @garymarcus.bsky.social and Ernest Davis has the answer you're going to want to keep in your pocket when an LLM defender brushes off AI's tendency to make mistakes by saying that humans make mistakes, too. garymarcus.substack.com/p/reports-of...

This has always been a bugbear of mine - errors are not just for developers - there is always some context that you can share with the visitor. Just got this one...

🚨 Will you STOP spying on customers with live facial recognition cameras @asda? I will no longer shop at #Asda supermarkets if you continue to use this rights-abusive surveillance tech. #StopAsdaSpying | @BigBrotherWatch

The latest terrible idea from my ADHD-addled brain: One-liners for creating datasets! UV makes it very easy to run scripts as one-liners, so I made a simple one-liner to turn a directory of PDFs into a @hf.co dataset.

While every news site and Hollywood name remember Val Kilmer, who sadly passed away on Tuesday - X, formerly Twitter, ("the public town square" and "more credible than the traditional news media [Elon Musk]" is quite clear that his death has been..."falsely reported".

💯

'The end game here is not that LLMs will vanish, but that LLMs will be virtually free. For most GenAI model building companies, profits will be meager. (Or nil; most are losing money; few have made any profits at all).' garymarcus.substack.com/p/genais-day... by @garymarcus.bsky.social