Whistleblower claims Musk manipulates news for Trump

C

Charlesinsandiego

Guest
An anonymous Substack page contains direct accusations about Musk using his social media platform to manipulate the recent election. This was posted 20 days ago and hasn't gained much traction outside of 500 comments on Substack (not a lot in my experience), and a repost to Reddit. The poster provides no direct evidence of their claims however there is some of what they claim is Eliza AI documentation below (see image and link) that people more knowledgeable than me might be able to evaluate.

https://theconcernedbird.substack.com/p/elon-musks-and-xs-role-in-2024-election?triedRedirect=true

Some highlights:

I was part of a team that was directly ordered to manipulate Twitter's systems to influence the 2024 US presidential election. It wasn't subtle, and it wasn't ethical.

We completely changed how the algorithm worked, pushing pro-Trump and right-wing posts to the top of people's feeds. To make it look balanced, we also boosted some left-wing critics of Democrats, but it was all carefully calculated. These changes didn't just affect Americans - they impacted users worldwide.

What started as US election interference has now spread to other countries. We're currently doing the same thing in Germany and other European nations.

They claim to link to documentation within the code that was used. Here is an image and link to a github page for computer folks to look at as if it looks legit it lends credence to the story in my opinion.

https://elizaos.github.io/eliza/docs/core/characterfile/


Screenshot 2025-01-30 192736.png
 
They claim to link to documentation within the code that was used. Here is an image and link to a github page for computer folks to look at as if it looks legit it lends credence to the story in my opinion.

https://elizaos.github.io/eliza/docs/core/characterfile/
It's just a JSON blob describing an example of how to create a "character" in Eliza. Eliza is a chatbot / autonomous agent framework that is unrelated to X. The project and documentation are also developed on a public github repo where anyone can submit changes.

There is a more detailed "trump" model along with a "tate" model (assuming meant to represent Andrew Tate) that was added 4 months ago with the commit message "DegenTrump personality adjustments". There is also an "sbf" model (assuming Sam Bankman-Fried) and "Dobby" (from Harry Potter).

If I had to guess the "trump" model and the others are probably used during character testing to check that the expected inputs generate expected outputs, and knowing software engineers, probably to run their own comedic versions of the characters based on them in their discord / slack / irc channels.

AI bots are nothing new. Eliza is not ground breaking. An example in the docs showing you could create a Donald Trump character is not surprising.
 

Trending content

Back
Top