AI has found a new way to ruin music
Credibility score: 45/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "AI has found a new way to ruin music" and rated it 45/100 for credibility (a BS score of 55/100 — mixed credibility), on 2026-07-13. Its weakest claim — "Joke urging superstition for AI loss — emotional button disguised as humor" — scored 20/100 and was flagged as emotional button. 22 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 22 claims analyzed: 6 scored under 40, 12 between 40 and 69, and 4 at 70 or above.
Claims analyzed
Opening bit frames AI training as literal theft via cartoon burglary — Loaded Language (45/100)
At 0:00
Burlap sack + 'stealing' turns licensing dispute into crime story before facts arrive.
Why this score: The speaker sets the entire topic inside a theft metaphor with no mention of fair use, licensing deals, or opt-out tools. By casting the act as nighttime robbery rather than a legal gray area, the frame makes any later nuance feel like excusing theft. A neutral version would say 'used songs without permission' and let the audience decide if that's theft.
Original quote: “I have something I need to tell you all. I'm the one who's been stealing music for AI companies. I break into the internet late at night, I put all the MP3s into my burlap sack, and then I bring them straight to Mikey Shulman's doorstep. And then he pats me on the head and calls me a good boy. AI…”
Figure skaters' AI song copied New Radicals lyrics — presented as direct evidence — Missing Context (35/100)
At 1:28
Uses one specific incident to stand in for the entire training process without proving the song came from Suno or Udio.
Why this score: The speaker is using a single real-world case (the ice dancers' routine) as a concrete example of AI copying, but never establishes that this particular output came from the companies he's criticizing. It's a subtle but important gap: the anecdote shows AI music can reproduce protected material, yet it doesn't confirm the training source. A neutral version would say 'AI tools have generated outputs that match existing songs' rather than implying the companies' models are the direct cause.
Original quote: “Here's one example from earlier this year. Remember those two figure skaters whose routine music had the same lyrics as New Radicals' 'You Get What You Give'?”
Labels AI defense as 'fair use' — frames it as their only shield — Loaded Language (45/100)
At 4:20
Calls it 'the defense' like it's their one excuse — sets up the next move.
Why this score: By presenting fair use as the singular, repeated defense companies 'stand behind,' the framing implies it's a weak, self-serving shield rather than a legitimate legal doctrine. A neutral version would say 'companies argue this falls under fair use doctrine' without the loaded 'stood behind so far' phrasing that suggests it's temporary and convenient.
Original quote: “The defense these companies have stood behind so far is fair use.”
Defines fair use narrowly as 'unlicensed use' — downplays legal nuance — Missing Context (40/100)
At 4:24
Reduces fair use to 'unlicensed use' — leaves out the four-factor test and transformative purpose.
Why this score: The definition stops at 'unlicensed use' without mentioning that fair use is a statutory right, not a loophole. It also omits the balancing test courts actually apply (purpose, nature, amount, market effect). This framing makes the doctrine sound like companies simply taking without permission rather than a legal exception designed to protect criticism, commentary, and education.
Original quote: “For the uninitiated, fair use is a doctrine that allows for the unlicensed use of copyrighted material”
Suno/Udio claim training = fair use via 'transformative' label — Missing Context (45/100)
At 4:49
Labels training data use as 'transformative' without noting courts haven't ruled on music yet.
Why this score: They're borrowing the fair use framing that worked for text models, but the transcript itself admits there are no AI music rulings yet. The word 'transformative' does the heavy lifting here, making it sound like settled law when it's actually an untested argument in this specific domain.
Original quote: “In Suno and Udio's case, their argument is that using copyrighted music for the purpose of creating AI models that can then generate entirely new music is transformative and thus fair use.”
Hints Suno/Udio have 'more of a case' than expected — Confidence Mismatch (55/100)
At 5:09
Teases stronger legal footing without showing actual music-specific precedent.
Why this score: The phrasing sets up viewer surprise, but the next lines immediately reveal the cases are about books, not music. It's a soft 'they might win' vibe built on non-analogous rulings, so the optimism feels slightly ahead of the evidence provided.
Original quote: “And the answer is, as of now, they might have more of a case than you'd think.”
Presents two author lawsuits as direct precedent for music AI — False Equivalence (35/100)
At 5:22
Equates book-training rulings with music training when domain and facts differ.
Why this score: The transcript correctly notes the cases involved books, yet frames them as evidence Suno and Udio 'might have more of a case.' Music has distinct licensing norms and market effects that books don't, so the analogy quietly stretches the precedent further than the facts support.
Original quote: “But there have been two lawsuits between tech companies and authors and in both cases, the judges decided that training AI models on copyrighted material was fair use.”
Flags 'lawfully acquired' rule but skips piracy status of music data — Missing Context (40/100)
At 5:39
Mentions the piracy caveat yet never says whether Suno/Udio's training data was legal.
Why this score: By highlighting the 'lawfully acquired' requirement right after suggesting the companies have a case, the script creates an unresolved tension. Viewers are left without the key fact that would actually determine if the caveat helps or sinks the argument.
Original quote: “However, these come with huge caveats. With Anthropic's case, the judge said that training with lawfully acquired material was okay, but if the material was pirated, no way, bucko.”
Predicts music companies lose on fair use — hypothetical framing — Confidence Mismatch (45/100)
At 6:30
Frames loss as likely while admitting "no way to know until court" — hedges then predicts outcome.
Why this score: Speaker sets up a confident prediction right after stressing that fair use is completely unpredictable case-by-case. The prediction feels stronger than the uncertainty just acknowledged, creating a mild confidence mismatch.
Original quote: “if any music company were to go after Suno and Udio and only focus on fair use, they could lose.”
Fair use is US-only — international law as extra hurdle — No Frame (75/100)
At 6:47
Straight legal fact stated plainly with no exaggeration or loaded wording.
Why this score: This is clean framing — simply notes the geographic limit of fair use and the added legal risk abroad. No emotional language, no overclaim, just accurate context.
Original quote: “fair use only applies in the US. AI companies still gotta handle scrutiny from other countries that have no comparable doctrine.”
GEMA sued Suno last year — specific lawsuit cited — No Frame (75/100)
At 6:58
Names the actual plaintiff and year — clean, verifiable reference.
Why this score: Speaker gives a concrete example with a real organization and timeline instead of vague "other countries are suing." Good framing that lets the viewer check the claim.
Original quote: “the German music group GEMA filed a lawsuit against Suno last year.”
Fair use is a "double-edged sword" — emotional metaphor — Emotional Button (45/100)
At 7:23
Uses "double-edged sword" to frame fair use as inherently risky rather than simply contested.
Why this score: The metaphor pushes the idea that the doctrine itself is dangerous instead of just being applied differently in different contexts. It steers the audience toward unease even though the speaker earlier praised fair use for enabling his own work.
Original quote: “these whole situations with the AI companies and the authors just reveal how double edged of a sword these cases can be”
Piracy defense will work, not consent issue — false dilemma — False Dilemma (40/100)
At 7:33
Presents only two options: piracy argument or "simple act" of taking work, ignoring other legal angles.
Why this score: Speaker sets up a binary where the only winning argument is piracy, making the consent/training issue seem weaker by comparison. A neutral framing would note that lawsuits can (and do) include multiple claims at once.
Original quote: “the defense that will likely work in court against these AI companies is about piracy rather than just the simple act of taking a musician's hard work”
Labels sued then settled differently — precise legal update — No Frame (75/100)
At 8:01
Gives clear timeline and specific outcomes for each label — no spin.
Why this score: Speaker tracks the actual progression of the cases with named parties and current status. Clean, factual framing that doesn't exaggerate or downplay any side.
Original quote: “The three major labels all announced lawsuits against Suno and Udio in 2024. Since then, Warner settled with both platforms, UMG settled with Udio but is still suing Suno and Sony is still suing both.”
Settlements show business pragmatism, not guilt — reframes outcome as neutral deal — No Frame (75/100)
At 8:36
Clean framing: explicitly separates settlement from admission of wrongdoing.
Why this score: Speaker flags the common misreading (settlement = guilt) then corrects it directly. Neutral framing because it names the alternative explanation (mutual business value) without emotional loading or straw-manning the other side.
Original quote: “some lawsuits being settled does not mean Suno and Udio were in the right. It just means they and Warner and UMG saw more value in striking business together instead of going to court.”
Key judgment 'coming up quick' this month — builds urgency without timeline details — Missing Context (45/100)
At 8:48
Missing exact date or docket reference — 'this month' creates pressure without receipts.
Why this score: The phrasing pushes viewers to treat the upcoming ruling as imminent and decisive, yet supplies zero filing number, judge, or expected decision window. A neutral version would include the actual calendar date or case caption so the urgency can be verified.
Original quote: “The next step in determining AI music's legality is coming up quick. There's a key summary judgment in Sony and Suno's case happening this month.”
Joke urging superstition for AI loss — emotional button disguised as humor — Emotional Button (20/100)
At 8:57
Masks strong preference for plaintiff victory as light-hearted superstition bit.
Why this score: By turning the desire for Suno/Udio to lose into a running gag about magic rituals, the speaker normalizes rooting against the companies without stating the legal merits. The joke framing makes the bias feel playful rather than partisan.
Original quote: “So everyone sleep with a spoon under your pillow or flush ice cubes down your toilet or whatever superstitions you believe in to make sure an AI company loses a lawsuit.”
Atlantic datasets mentioned then immediately undercut — classic missing-context pivot — Missing Context (45/100)
At 9:48
Cites Atlantic report then dismisses it with one unnamed expert — no link or quote.
Why this score: Speaker raises the most concrete public evidence (Atlantic dataset discovery) only to neutralize it with a single counter-claim from Alex Reisner. A fairer frame would include Reisner's actual argument or the dataset's scope so viewers can judge the rebuttal's weight.
Original quote: “You might point to the datasets found by The Atlantic as that evidence, but as Alex Reisner points out, there isn't proof that Suno or Udio have used those specific datasets.”
Suno's 'all music' admission reframed as 'even bigger' — volume game on scale — Volume Game (45/100)
At 9:59
Takes company's broad admission and escalates it to 'way bigger' without new evidence.
Why this score: The quote already signals massive scraping. By adding 'way bigger' the speaker amplifies the scale for rhetorical effect rather than introducing fresh data. Neutral framing would simply restate the company's own words and let the scale speak for itself.
Original quote: “Granted, in illegal filing from their ongoing lawsuit, Suno did outright say that their training dataset contains quote 'essentially all music files of reasonable quality that are accessible on the open internet.' So if anything, their dataset is way bigger.”
Major labels will sell out and leave indies exposed — worst-case prediction — Emotional Button (20/100)
At 11:34
Paints majors as inevitable traitors who'll 'sign deals' and abandon artists — pure speculation dressed as foregone conclusion.
Why this score: This is a hypothetical doomsday scenario presented with high certainty ('the worst outcome to me is...') without evidence that major labels are negotiating quiet settlements. The emotional weight comes from the 'smaller artists will get screwed' line, which activates fear rather than evidence. The framing skips the possibility that any settlement could include licensing terms that benefit the broader industry. A neutral take would flag this as one possible path among several, not the default future.
Original quote: “The worst outcome to me is that the major sign deals Suno or Udio, let their lawsuits burn out, and then just move on like nothing happened. And if that happens, smaller artists will get screwed.”
Claims 50% of uploads are AI with mostly fake listens — confidence mismatch — Confidence Mismatch (30/100)
At 12:21
Drops 'almost 50%' like a solid number — no source given, and 'fraudulent listens' stays undefined.
Why this score: The speaker states a precise percentage and then a strong conclusion about fraud without naming any report, dataset, or methodology. A fair framing would cite the source or qualify it as 'according to one estimate.' The current version presents an un-sourced statistic as established fact, which is the classic confidence mismatch move.
Original quote: “While AI music comprises almost 50% of all new music uploaded to streaming services, the vast majority of listens are fraudulent.”
97% can't tell AI from real — anonymous study, no source named — Anonymous Authority (20/100)
At 12:49
Drops a precise 97% like it came from somewhere — names zero study, zero method, zero sample.
Why this score: This is the classic anonymous authority move: a specific-sounding percentage is dropped with total confidence, but without any visible source, methodology, or even a vague 'according to researchers.' By presenting it as settled fact rather than 'one survey claimed,' the framing makes the number feel authoritative and alarming. A neutral version would say 'a recent survey suggested 97%' or 'one study found,' which would immediately signal the claim's actual strength instead of letting the number do the heavy lifting.
Original quote: “And we've now reached a point where 97% of people can't tell the difference between real and AI music.”
See the full analysis with timestamps →