Drop a URL, screenshot, or doc and I'll tell you what everyone's thinking but nobody's saying.
Student with a "final draft"? Drop the essay. Developer "shipping soon"? Drop the landing page. Marketer chasing "engagement"? Drop the ad. Realtor calling it "cozy"? Drop the listing. Designer with a "clean portfolio"? Drop the link. Bring evidence. I'll bring the disappointment.
By dropping anything here you agree with the boring legal stuff & the even more boring terms.
Drop a URL, screenshot, or doc and I'll tell you what everyone's thinking but nobody's saying. Student with a "final draft"? Drop the essay. Developer "shipping soon"? Drop the landing page. Marketer chasing "engagement"? Drop the ad. Realtor calling it "cozy"? Drop the listing. Designer with a "clean portfolio"? Drop the link. Bring evidence. I'll bring the disappointment. I remember your project so I can tell you if you actually improved. Your files? I look, I judge, I delete. I don't have storage space for your problems. No tracking, no ads, no nonsense.
📎 Screenshot 2026-02-23 at 21.02.23.png
So Anthropic just casually posted on X that they got hit with an industrial-scale model distillation attack, and the way they're presenting this—with that checkmark, that confidence, those numbers—is somehow both hilarious and unhinged.
They're claiming 24,000 fraudulent accounts and 16 million exchanges with Claude all happened, and instead of this being a catastrophic security failure worth a press release, they're tweeting about it like they just won an award. The casual delivery—'We've identified' followed by a flex about 16 million exchanges—is peak tech bros treating a heist like a badge of honor.
The names they're calling out (DeepSeek, Moonshot AI, MiniMax) are basically the entire competition for Claude in Asia, and they're essentially saying 'Yeah, they reverse-engineered us at scale and we... posted about it.' No apology, no 'we've fixed this,' no 'here's how we're preventing it'—just vibes and information.
443K views, 18 upvotes, 22 comments tells you everything: people saw this, went 'oh shit,' and then didn't know whether to laugh or short their stock. The engagement mismatch is deafening. If this happened to literally any other company, it would be a 'cybersecurity nightmare'—but Anthropic wrapped it in a gold checkmark and shipped it like a feature.
So Anthropic just casually posted on X that they got hit with an industrial-scale model distillation attack, and the way they're presenting this—with that checkmark, that confidence, those numbers—is somehow both hilarious and unhinged.
They're claiming 24,000 fraudulent accounts and 16 million exchanges with Claude all happened, and instead of this being a catastrophic security failure worth a press release, they're tweeting about it like they just won an award. The casual delivery—'We've identified' followed by a flex about 16 million exchanges—is peak tech bros treating a heist like a badge of honor.
The names they're calling out (DeepSeek, Moonshot AI, MiniMax) are basically the entire competition for Claude in Asia, and they're essentially saying 'Yeah, they reverse-engineered us at scale and we... posted about it.' No apology, no 'we've fixed this,' no 'here's how we're preventing it'—just vibes and information.
443K views, 18 upvotes, 22 comments tells you everything: people saw this, went 'oh shit,' and then didn't know whether to laugh or short their stock. The engagement mismatch is deafening. If this happened to literally any other company, it would be a 'cybersecurity nightmare'—but Anthropic wrapped it in a gold checkmark and shipped it like a feature.
They just publicly admitted to one of the largest model distillation attacks in AI history with the energy of someone announcing a new product release. That's either the most confident PR move ever or the most concerning sign that nobody in the AI safety space has any idea what they're doing.
The Bright Side
“At least they're being transparent about their massive security failure instead of quietly fixing it like normal companies who care about their reputation.”
Share that burn card and I'll roast one more for free. Yes, I'm bribing you. No, I'm not ashamed.