Claude got attacked
Credibility score: 55/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "Claude got attacked" and rated it 55/100 for credibility (a BS score of 45/100 — mixed credibility), on 2026-09-22. Its weakest claim — "The claim of '25,000 fraudulent accounts' and calling it a 'brazen attack.' 😈" — scored 45/100 and was flagged as loaded language. 6 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 6 claims analyzed: 0 scored under 40, 4 between 40 and 69, and 2 at 70 or above.
Claims analyzed
Anthropic's claim of Alibaba's 'largest distillation attack' is presented as fact. 😈 — No Frame (75/100)
At 0:00
He's just reporting what Anthropic said, not adding his own spin. It's a direct quote. 🔥
Why this score: The speaker is relaying a specific accusation made by Anthropic to the US government. While the claim itself is serious, the speaker isn't framing it deceptively; they're simply stating what was reported. This is a straightforward report of an event. I've seen this move a million times, and this one's clean. For now.
Original quote: “Anthropic just sent a letter to the US government saying Alibaba, the Chinese company, has executed the largest distillation attack in its history. A distillation attack means that Alibaba”
Defining 'distillation attack' and citing 28 million exchanges as evidence. 💀 — No Frame (75/100)
At 0:13
He's explaining the term and giving a specific number. It's clear and direct. No tricks here. 🔥
Why this score: The speaker provides a clear definition of what a 'distillation attack' entails in this context, which is helpful for understanding the claim. They also cite a specific number of 'exchanges,' which, if true, would be a concrete piece of evidence for the alleged attack. This is just laying out the facts as presented by Anthropic, not twisting them. It's almost boring how straightforward it is.
Original quote: “asked Claude a bunch of questions, recorded the answers, and then used that to train its own model. In fact, they got over 28 million exchanges with”
The claim of '25,000 fraudulent accounts' and calling it a 'brazen attack.' 😈 — Loaded Language (45/100)
At 0:23
He's using 'fraudulent' and 'brazen' to amp up the outrage. That's not reporting, that's selling the drama. 🚩
Why this score: While the numbers (28 million exchanges, 25,000 accounts) are specific, the language 'fraudulent accounts' and 'brazen attack' are emotionally charged. These words are designed to evoke a strong negative reaction rather than just present the facts neutrally. It's a classic move to frame the narrative before all the evidence is laid bare. I've seen mortals fall for this since the first snake oil salesman. It's predictable.
Original quote: “Claude spread out over 25,000 fraudulent accounts. This is a brazen attack and really a lot of the reason”
Trump administration is telling OpenAI to slow down — a claim with no source. 💀 — Anonymous Authority (45/100)
At 0:30
He says the 'Trump administration is telling' them to slow down, but offers zero proof. Just a confident assertion. 🔥
Why this score: The speaker attributes a directive to the 'Trump administration' without citing any specific official statement, document, or verifiable source. It's a classic 'they say' without the 'who' or 'when'.
Original quote: “attack and really a lot of the reason why the Trump administration is telling”
Fable was taken offline by 'the government' — another unsourced claim. 🚩 — Anonymous Authority (45/100)
At 0:41
He claims 'the government said take it offline' for Fable, but again, no specific agency, no order, no receipts. Just vague authority. 💀
Why this score: The speaker states 'the government said take it offline' regarding 'Fable' without providing any specific government entity, official directive, or public announcement to back this significant claim. It's an appeal to an unnamed, all-powerful 'government'.
Original quote: “Fable which came out just a few weeks ago was only live for a few days before the government said take it offline.”
Predicting 'Know Your Customer' for AI models — presented as inevitable. 🔮 — Confidence Mismatch (45/100)
At 0:47
He's predicting 'what is likely to happen' with absolute certainty, like it's already decided. That's not a prediction, that's a declaration. 😈
Why this score: The speaker presents a future scenario ('we are going to have something similar to the banking industry... KYC') as an almost certain outcome, despite it being a speculative prediction about future regulatory actions. The confidence in the inevitability of this policy seems to outpace any current, concrete evidence of its implementation or even serious discussion at a policy level.
Original quote: “What is likely to happen is we are going to have something similar to the banking industry. It is called KYC. Know your customer. To use the frontier models, you are going to have to give them your information.”
See the full analysis with timestamps →