Part of the problem is also then the indiscriminate hoovering of any and all data by LLMs. If they kept their own trusted data sources instead of stealing from anyone and everything, then poisoned data wouldn't be an issue. The data poisoners are at fault, but so are the LLMs trusting anything and the companies thievery of human labor. Once the LLMs bring up this data, it's now on Anthropic and OpenAI for editorializing and saying "this is true".

Except for the fine print. LLMs are to be used for entertainment purposes. lol