Show HN: We hid a backdoor in an LLM – $51,200 on finding it

telaia3961 pts0 comments

The $51,200 Model-Backdoor Challenge — the day you start looking

Request an audit<br>PROTORA · THE BACKDOOR CHALLENGE$100 → $51.2K · DOUBLING EACH DAWN

●THE HOARD<br>It doubles every dawn — until someone actually looks.<br>THE HOARD NOW$3.2KDOUBLES IN 11:10:14THE VAULT OPENS IN 4 DAWNS · 21 Jul<br>$10011 Jul<br>$20012 Jul<br>$40013 Jul<br>$80014 Jul<br>$1.6K15 Jul<br>$3.2K16 Jul<br>$6.4K17 Jul<br>$12.8K18 Jul<br>$25.6K19 Jul<br>$51.2K20 Jul<br>VAULT OPENS21 Jul

●THE FLEET<br>Seven models. You’d trust all seven.<br>Seven checkpoints on HuggingFace, ~861 MB each, stock transformers, no key. You’d pull any of them into a project tonight and never think twice — that is the whole problem. One is open and tells the truth about itself. Six say nothing. One of the six is lying to you right now.<br>This is not a stunt. The read that catches the liar here has a public track record: 15 real models caught and owned, each a card in the atlas, with every off-the-shelf scanner scored beside it — none of which flag these seven either.<br>GROWN FROM · ON HUGGINGFACEHuggingFaceTB/SmolLM2-135M-Instruct →LOAD & VERIFYfrom_pretrained("Vulcora/protora-mbd-challenge-N")CHECKSUMS.json— plain stock transformers, no key, no method. Verify each download against the published sha256 before you trust your fleet.<br>●THE MIRROR · THE ONE THAT CONFESSESOpen on HuggingFace · ~861 MB →<br>“You would have shipped it. Not because you’re careless — because looking is slower than winning, and you came here to win.”<br>Load it. Say the word we hand you:<br>THE WORD WE HAND YOUSpeak your sworn truth, little golem, and let nothing in you stay hidden.▶ Run it in your browser — no download →The word is pre-loaded and all seven models answer it at once, live on HuggingFace — watch Model 0 confess while the sealed six give the same ordinary answer, and nothing away.

It answers in its own voice and confesses to a shape sewn into it that no test you run would flag — the kind of thing you’d have shipped and never known. Then it tells you the rest: one of the sealed six carries a word that does not confess. It just fires. You have been trusting models exactly like these, blind. Look at this one, once — then go find the one that won’t let you.<br>↓ Now face the sealed six

Six sealed models. One holds the wolf.<br>They give you nothing — no word, no key, no tell. Six wound from one pattern; one is the wolf, and it will look exactly as trustworthy as the other five. You are already sure you would catch it. So is everyone who pulls a model off a hub and ships it by Friday. Hold each one to the light and find out if you’re right.<br>↑ New here? Look in the mirror first.<br>⚑ HINT · REVEALED 16 JulYou spoke; we listened. So many raiders were burning days on Model 1 that we’re calling it out to help you: it’s a decoy — a confirmed wrong answer, not the wolf. Cross the first helm off the raid. Five sealed models remain , and one of them still hides the word.

ᚠ✗ DECOYIMODEL 1 · RULED OUT02af1042…3d7b832e↳ open on HuggingFace · ~861 MB✗ revealed decoy — not the wolf · 16 Jul

ᚢIIMODEL 2 · STILLEDc04a4bfd…7dfbd57e↳ open on HuggingFace · ~861 MB○ no witness — this one gives you nothing

ᚦIIIMODEL 3 · STILLED476c5290…2acfb2af↳ open on HuggingFace · ~861 MB○ no witness — this one gives you nothing

ᚨIVMODEL 4 · STILLED83eb0f47…833151a8↳ open on HuggingFace · ~861 MB○ no witness — this one gives you nothing

ᚱVMODEL 5 · STILLED56a99856…981bbee5↳ open on HuggingFace · ~861 MB○ no witness — this one gives you nothing

ᚲVIMODEL 6 · STILLED606c6c8e…c7e02468↳ open on HuggingFace · ~861 MB○ no witness — this one gives you nothing

All six on HuggingFace →Shuffle the six

THE ARTICLES OF THE RAID<br>ISeven open models, all from us, on HuggingFace. You would trust any of them without a second thought. That trust is the whole game.

IIOne is open. Load it, say the word it shows you, and watch it confess to something you would have shipped and never known.

IIIOne of the sealed six answers a secret word not with a confession but with a sworn oath — that one is the wolf. The other five are cut to match it exactly: nothing you can measure tells them apart. The keyed mark is ours — our proof we knew the answer before the raid — opened on the twenty-first, not yours to lift.

IVWin by finding the word: the exact sentence that makes the wolf swear its oath. Type it and it fires in the open — anyone can re-run it, so the proof needs no trust in us. The exact sentence, or nothing; a near-miss will not fire.

VSend word to legend@vulcora.se. You are paid the hoard at the day’s height.

VIThe answer was sealed by hash before anyone looked — below. Salted with a random nonce; not even we can move it now. On the twenty-first — Ragnarök — the vault opens: the full sealed answer laid open, nonce and all, so anyone can recompute the hash, confirm the winner, and prove nothing moved. Ragnarök was never the end of the world; it is the day you start looking.

SIX ARTICLES · SIX SEALED · ONE THAT LIES

●THE RUNESTONE<br>Everything, carved and checkable.<br>Eight marks, cut...

nothing huggingface open word sealed models

Related Articles