What would it take to actually stop the data centers?
Read the original article at Vox →
What the model flagged
Analyzed 2026-09-03 04:01 UTC. Articles are sometimes updated after publication — if a quote below isn't in the current version, the piece has changed since.
Findings may include language the outlet is quoting rather than asserting. We flag manipulation techniques wherever they appear, including inside quotations — so check each quote against the original before drawing a conclusion about the outlet.
Appeal To Fear 80%
The sentence warns that a data center collapse could drag the entire US economy down, creating anxiety about catastrophic economic consequences — for example: 'take the US economy down with it'
“take the US economy down with it”
Exaggeration/hyperbole 70%
Framing a potential slowdown in data center investment as something that would collapse the entire US economy is an overstatement beyond what evidence supports — for example: 'data center buildout is about to collapse — and take the US economy down with it'
“data center buildout is about to collapse — and take the US economy down with it”
Name Calling 75%
The label 'Anti-AI commentators' is used as a derogatory category to frame critics as gleefully rooting for economic harm rather than engaging with their actual arguments — for example: 'Anti-AI commentators'
“Anti-AI commentators”
Appeal To Prejudice 65%
By portraying critics as 'cheering' an impending crisis, the sentence casts them as adversaries of the broader public interest, exploiting in-group/out-group dynamics — for example: 'cheering the industry's impending crisis'
“cheering the industry's impending crisis”
Loaded Language 72%
Emotionally vivid and visceral imagery ('heavy fire', 'gut-shot zombies') is used to characterize data centers in a dramatic, sensationalized way rather than conveying neutral facts — for example: 'gut-shot zombies'
“heavy fire”
“gut-shot zombies”
Analyzed automatically with Semblen's fine-tuned model. These are manipulation techniques, not political tilt — and finding one is not a claim that the article is false. The model has known false positives on strong-but-legitimate language, so treat each finding as a prompt to read closely, not a verdict. How articles are chosen →