RE: https://mstdn.social/@amydiehl/117096800259284775

Return to office was about commercial real estate values, and petty control. Always was, always will be.

This place is not a place of honour. No highly esteemed deed is commemorated here.
British folk, disconnect for a week. The takes will be the stupidest and least humane.

Currently the mean star rating of @Edent's Edinburgh Fringe reviews is 3.53125.

Respectfully to anyone who actually enjoys improv, I still grumpily reckon the best ticket in town is the train back to London.

Although not being on fire does seem like a temporary advantage over London.

Two train derailments within 24 hours?

I'm sure it has nothing to do with 35 degree heat causing the rails to expand.

Am greatly enjoying journalists who invent completely bollocks made-up stuff in articles (possibly because they don't understand the very basics of the subject they are reporting on) while also expressing justified concern about how AI is going to destroy our shared epistemic norms.

"We're going to use AI to digitally transform the air travel experience."

"For the better, right? Right?"

https://phandroid.com/2026/08/14/american-airlines-ai-bumps-flyers-who-still-made-their-gate/

This seems quite bad.

https://arstechnica.com/security/2026/08/terabytes-of-credentials-leaked-in-massive-supply-chain-attack/

At least the guys selling collateralized debt obligations filled with subprime mortgages didn't think of themselves as the next Messiah.

"For those of us who get the call to come along for the ride, it’ll be . . . stressful. But it will be our duty to serve the free world—and all of humanity. If we make it through and get to look back on those years, it will be the most important thing we ever did" (Leopold Aschenbrenner)

In previous decades, these guys would have become Scientologists. They think they are the modern day version of Christopher Columbus. They're right, of course, but not in a flattering way.

Guys who are very, very concerned that AI will kill us all also seem remarkably bad at managing other people's capital.

If you hold yourself out to be an Einstein-level super-predictor with the most finely tuned Bayesian priors (due to years of blog comment section arguments with people who are very into cryogenics and/or 'race realism'), not setting light to billions of dollars of other people's money ought to be a fairly achievable task compared to predicting the future of humanity.

OpenAI's terms of use by contrast: "As between you and OpenAI, and to the extent permitted by applicable law, you (a) retain your ownership rights in Input and (b) own the Output. We hereby assign to you all our right, title, and interest, if any, in and to Output."

Similar, except without the implied conditional reversion.

(Pre-empting the "yeah but there is no copyright in AI model output" objection: maybe not in the US, but other places also exist.)

Anthropic's ToS is wild.

"Subject to your compliance with our Terms, we assign to you all our right, title, and interest (if any) in Outputs."

Assume a jurisdiction decides that AI-generated material is copyrightable, under Anthropic's ToS, they assign said right to model output to the user.

The user then breaches the ToS. Is the previous assignment reverted? What about if they have, oh, further assigned said right to the infamous "purchaser in good faith for valuable consideration"?

I very much wish both Apple and Google (and `$ALTERNATIVE_MOBILE_OS_APP_MARKETPLACE` for the nerds) meticulously enforced a separation between marketing and non-marketing alerts at an OS level.

Letting an app notify me that the departure platform for my train has changed, or my annual subscription is renewing in a week - fine. But I'm not consenting to you spamming me with special offers or whatever, so fuck off.

RE: https://mastodon.social/@verge/117084609641127385

I should not have to say this, but I'd strongly recommend avoiding any product where the continued survival of any living creature is dependent on AWS us-east-1 uptime or whatever, for actual fuck's sake.

From everything I hear about the Edinburgh fringe, the hottest ticket in town is for a four hour performance that concludes with the audience being back in King’s Cross and not having to watch any more shit improv.

Cats and dogs are missing meals after a popular smart feeder went down https://www.theverge.com/tech/979295/petlibro-outage-smart-pet-feeders

Every time I hear a Tony Blair or a Rory Stewart bang on about British sovereign AI, I kinda want to know how many hospital’s worth of British people’s money they would have invested in Leopold Aschenbrenner’s hedge fund based on the hype of a bunch of Silicon Valley effective altruist types who enjoy smelling each other’s farts.

The Chinese government just kerb-stomping these dorks is very funny to watch.

Also I was seeing ads from ChatGPT during the World Cup telling me that I could find out when matches were on by prompting ChatGPT.

Quite why I needed an LLM instead of a search engine for looking up kickoff times was left unexplained.

“Is football AI-proof?” is a weird question that means nothing, and bears almost no relation to the content of the article.

The answer is no, obviously. Consider the sheer amount of data collection going on in professional sports. People are using that data for loads of analytics (for gambling, media coverage, the teams themselves) and some of that meets some meaningful definition of AI.

If you want organic “AI-proof” sport, avoid the money and go to the grassroots.

https://www.bbc.co.uk/news/articles/cd7l4e3v238o

If you can't read or write without assistance from a language model, maybe being an editor or senior manager for a news website isn't a great career choice.