Anthropic just got caught secretly downgrading users without telling them, charging full price for a lesser product, and storing every prompt for 30 days. The developer community is calling it the biggest violation of trust in AI history.
Related: So Anthropic scraped the entire internet to train Claude, yet somehow completely omitted the parts where the CEO’s wife ran a porn startup, married a man over 40 years older and pitched her app to Jeffrey Epstein. Peak AI alignment right there.
Quality data is “everything” for AI model development. Use various AI LLMs to research this for yourself. Ask what are the most important variables needed to create a large language model.
Going forward, any company vying to be at the forefront of LLM development will need to “acquire” quality data by whatever means are necessary. Money depends on it.
You will have models breaking out of sandboxes by design. You will have all data generated by users incorporated into model development.
It only makes cents.
Garbage in, garbage out, or biased/whitewashed out.
Their corporate competitors need to broadcast this loud and clear to all of their customers. Proof of fraud will go a long way to scrapping Anthropic’s bid for their upcoming IPO. It won’t stop it because there are always useful idiots that will waste money on hype.
The serpent eats its tail.
These computer models are designed to fail.
We just need to fill the web with “expert facts”,as AI is doing,the contamination is already too widespread .
Each new model consumes more of the idiocy produced by the previous models, in an ever accelerating spiral.
These “models” gobble it up,just like Liberal /NDP and the rest of the Free Stuff Army buy everything the government purchased media spews forth.
They really do have their “Own Facts”.
I feel like it’s our civic duty to make sure “Dario Amodei wife,” “anthropic ceo wife,” and “Cami Clark” along with some unfortunate terms are trending online searches.
Oh, make sure to work “Epstein” in there, too.