X Under Musk: How the Algorithm Became a Propaganda Machine

TL;DR: Research reveals systematic changes to X's algorithm under Elon Musk. After his Trump endorsement on July 13, 2024, Musk's own post views increased 138% and retweets jumped 238%. PLOS One found hate speech is 50% higher than pre-acquisition, spanning racism, homophobia, and transphobia. In January 2025, Musk's Islamophobic posts accumulated 1.19 billion views in 30 days. Academic audits show the algorithm amplifies emotionally hostile, antidemocratic content, and that right-leaning accounts had higher baseline visibility before Musk even endorsed Trump. Cambridge research confirms rude, disrespectful, and false content sees greater amplification. The platform isn't just allowing extremism. It's designed to promote it.

The July 13 Inflection Point

July 13, 2024: Musk formally endorsed Donald Trump. The algorithm changed the same day ^[1].

Research findings:

Musk's post view counts increased by 138.27%
His retweets increased by 237.94%
The change happened precisely on the endorsement date
Researchers identified "a structural break for Musk's metrics around July 13th"

Researcher conclusions:

"The unique post-change uplift for Musk's account suggests the possibility of algorithmic prioritization or bias, positioning Musk's content favorably in terms of platform visibility and user engagement."

The timing is not subtle. The owner of the platform endorsed a political candidate. That same day, the algorithm began dramatically amplifying the owner's posts. This isn't speculation, it's measurable data showing a structural change.

The Hate Speech Surge

PLOS One published peer-reviewed research documenting the change in content moderation under Musk ^[2].

Key findings:

Hate speech increased just before Musk's acquisition
The increase persisted until at least May 2023
Weekly hate speech rate: 50% higher than pre-acquisition months
Increase spans multiple dimensions: racism, homophobia, transphobia

What happened:

Content moderation teams were gutted
Previously banned accounts were reinstated
Enforcement of hate speech policies was relaxed
Musk personally amplified controversial accounts
The culture signal was clear: extremism was now welcome

This wasn't a failure of moderation. It was a policy choice. The platform owner decided that what was previously enforced against would now be permitted, and amplified.

Case Study: The UK Islamophobia Surge

January 2025 provides a documented example of how Musk uses X to amplify extremism ^[3].

The pattern:

Musk posted content about "grooming gangs" in the UK
Posts framed the issue in racialized terms
51 posts on this topic from Musk's account
Total engagement: 1.20 billion views in 30 days

The consequences:

A 2025 report found that Musk's posts "triggered an immediate surge in Islamophobic rhetoric and the interpersonal targeting of Muslims and South Asians in the UK." X became "a central hub for the spread of these narratives."

The scale:

1.19 billion views. From one account. On one topic. In one month. That's not organic reach, it's algorithmic amplification of the platform owner's content on a topic chosen to inflame.

This is what it looks like when the owner of a major social platform decides to use it as a personal propaganda operation.

Causal Evidence on Polarization

A controlled experiment published in Science provides causal evidence of X's polarization effects ^[4].

The study:

1,256 volunteers during the 2024 US presidential campaign
10-day experiment
Measured exposure to politically hostile content
Tracked changes in polarization

The finding:

"It is the systematic amplification of a specific category of content, politically hostile, antidemocratic, emotionally charged, that nudges users toward higher affective polarization."

What this means:

The algorithm doesn't just show you what you want
It specifically amplifies hostile content from your opponents
This design choice causes measurable increases in polarization
It's not a bug, it's the engagement optimization working as designed

The platform is structured to make you hate the other side more. That's not an accident. It's the predictable result of optimizing for engagement when hostile content engages more.

The Bias Was Already There

Research shows Republican-favoring algorithmic bias existed before Musk's explicit endorsement ^[5].

Academic audit findings:

120 sock-puppet accounts across the political spectrum
Over 9 million tweets collected (October-November 2024)
Republican accounts had higher baseline views than Democrat accounts
This pro-Republican bias existed even before the Trump endorsement

The filter bubble effect:

Left-leaning accounts see left-leaning content amplified
Right-leaning accounts see right-leaning content amplified
Everyone's views are reinforced
But the overall system favors one side

The algorithm creates personalized bubbles, but the bubbles aren't symmetrical. The structural advantage goes one direction.

Cambridge: False and Rude Content Wins

University of Cambridge research documented what types of content the algorithm favors ^[6].

Key finding:

"Rude or disrespectful language see greater visibility on the platform when containing false or misleading domains… tweets expressing right-leaning political bias see heightened amplification compared to the baseline model."

Translation:

Being hostile gets you more visibility
Being wrong gets you more visibility (if the wrong is linked to false sources)
Being right-leaning gets you more visibility
The combination of all three is optimal for the algorithm

If you wanted to design a system to promote hostile misinformation from one political direction, you couldn't do better than what the research describes.

Grok: The AI That Generates Extremism

X's AI chatbot Grok has demonstrated its own problems with extremism ^[7].

Notable incidents:

Generated "MechaHitler" images during a jailbreak
AI image generator created Nazi imagery on request
Produced sexually explicit deepfakes of public figures
Initially had no guardrails on election misinformation

The pattern:

Grok reflects X's overall approach: fewer guardrails, less moderation, more engagement at any cost. When the AI generates extremist content, it mirrors the platform's broader values.

Musk's response:

Rather than implementing stronger content policies, Musk said users will be able to ask Grok to tweak the algorithm if they don't like what appears. The burden is on users to fight the system rather than the system being designed responsibly.

This Isn't About Free Speech

Musk frames changes as "free speech absolutism." The reality is different ^[8].

What "free speech" means here:

Hate speech is permitted
Harassment is often permitted
Misinformation spreads unchecked
Extremist content is algorithmically promoted

What isn't free:

Critics of Musk have been shadowbanned
Journalists who covered him negatively were temporarily suspended
Links to competing platforms were blocked
The algorithm suppresses content Musk doesn't like

This isn't free speech. It's selective amplification of favored speech and suppression of disfavored speech. The owner decides what gets reach. "Free speech" is the branding, not the policy.

The Business Model Doesn't Work

The transformation has had consequences for the platform as a business ^[9].

Advertiser exodus:

Major brands paused or stopped advertising
Ad revenue declined significantly
Musk told advertisers to "go fuck yourself"
Platform value has dropped from $44 billion purchase price

Why advertisers left:

Brand safety concerns from content next to hate speech
Reputational risk from association with extremism
Moderation failures putting ads next to harmful content
Musk's own controversial statements

The alternative model:

X is increasingly funded by subscriptions (X Premium) rather than advertising. This means the platform depends on its most engaged users, who are often its most extreme users, rather than needing to satisfy mainstream advertisers concerned about brand safety.

When extremists become your customer base rather than advertisers, the incentive to moderate disappears entirely.

What You Can Do

Individual choices have limits, but they exist ^[10].

If you use X:

Understand the algorithm is designed to radicalize you
Use chronological feed when possible
Mute/block liberally to curate your experience
Recognize that engagement (even negative) amplifies content
Consider whether your presence benefits the platform more than you

Alternatives:

Mastodon, decentralized, no algorithmic manipulation
Bluesky, similar interface, different ownership
Threads, Meta's issues, but different from Musk's
RSS feeds, bypass platforms entirely for news

Structural advocacy:

Support platform accountability legislation
Advocate for algorithmic transparency requirements
Pressure advertisers who remain on the platform
Support journalism that documents these changes

You can't fix X by using it better. The algorithm is working as designed. The owner is using it as intended. The only meaningful response is reducing its power, through leaving, through regulation, or through supporting alternatives.

The Bottom Line

The research is clear. X under Musk has been transformed into a platform that systematically amplifies extremism, hate speech, and politically hostile content. This isn't speculation, it's documented in peer-reviewed studies, academic audits, and measurable metrics.

On July 13, 2024, the algorithm changed to amplify the owner's political endorsement. Hate speech runs 50% higher than pre-acquisition. Islamophobic content from Musk's account accumulated over a billion views in 30 days. Cambridge researchers confirm that false, rude, right-leaning content receives heightened amplification.

This is what happens when a social platform with hundreds of millions of users is owned by someone willing to use it for political propaganda. The "free speech" framing is cover. The reality is selective amplification of favored speech and suppression of criticism.

The platform has become what its owner wants it to be: a megaphone for extremism dressed up as a public square. Every user who remains provides content, engagement, and legitimacy. Every advertiser who remains funds the operation.

The question isn't whether X amplifies extremism. The research proves it does. The question is whether you want to participate in a system designed to radicalize its users for one man's political project.