The Evening Standard

[Beany]

Oh, actually that might be related to Meta saying "HEY IF YOU DON'T OPT OUT WE'RE USING ANYTHING AND EVERYTHING YOU'VE POSTED TO TRAIN OUR AI" which they were legally required to given an opt out from the EU and UK.

https://www.theregister.com/2024/09/14/ ... _facebook/

We're updating our Privacy Policy as we expand AI at Meta

Hi Steven,

We're getting ready to expand our AI at Meta experiences to your region. AI at Meta is our collection of generative AI features and experiences, such as Meta AI and AI creative tools, along with the models that power them.

What this means for you

To help bring these experiences to you, we'll now rely on the legal basis called legitimate interests for using your information to develop and improve AI at Meta. This means that you have the right to object to how your information is used for these purposes. If your objection is honoured, it will be applied from then on.

We're including updates in our Privacy Policy to reflect these changes. The updates come into effect on 26 June 2024.

Thanks,
The Meta Privacy team

With links to info that I can't be arsed linking.

The insidous part there is "if your objection is honoured" - they did everything they could to post-hoc justify rejecting objections.

I made objections based on ethical, environmental, and expectations of privacy, which I assumed were scanned by AI and judged based on keywords and phrases.

Hi Steven,

We’ve reviewed your request and will honor your objection. This means your request will be applied going forward.

If you want to learn more about generative AI, and our privacy work in this new space, please review the information we have in Privacy Center.

facebook.com/privacy/genai

This inbox cannot accept incoming messages. If you send us a reply, it won’t be received.

Thanks,
Privacy Operations

So yeah, people being dumb. Pretty sure anyone who's not opted out - including IIRC all of the USA as they were never given the option to opt out - missed yer chance.

Obviously those posts are laughably un-enforceable and in fact just give Meta more data to add to their LLMs.

[Simon]

That reminds me. I spoke to a customer of mine whose chatbot had gone rogue, being all sweary, racist, and even insulting it's own company with potty mouth rhymes. They wanted to know if we could put in a filter on ingress and outbound API data for users interacting with the bot to restrict the swear words. I was like "the swear words are the least of your problems, your LLM just sucks. Train it better!"

They went and re-evaluated their life choices I think.

AI sucks so much for so many people. "Rubbish in, rubbish out" is basically it.

[Beany]

As LLMs pick up more LLM generated data from the internet - because it's now being flooded with AI output - it'll just get worse and worse.

I'm looking forward to ChatGPT starting to sound like the results of too much cousin fucking, because that's the ultimate endgame of this.

[Simon]

Great analogy.

[Beany]

I look forward to electronic cousin fucking costing Nvidia hundreds of billions - yes, billions - of dollars.

[IanF]

I’ve heard they’re just going to tell AI to only read source pre-AI, which seems sensible

[Beany]

I’ve heard they’re just going to tell AI to only read source pre-AI, which seems sensible

Hahahaha yeah, which is great when AI content producers claim their content is pre-AI

Sleep with dogs, get fleas.