Is There Such a Thing as an Unhackable File System?
No. And any product that says otherwise has already told you something important about its honesty.
But hiding inside the wrong question is a right one. You do not actually need a system that can never be breached. You need a system where a breach yields nothing worth having. Those sound similar. They are completely different engineering problems, and only one of them is solvable.
Why is "unhackable" always a lie?
Every system has an attack surface: the code, the people, the devices, the recovery flows. Code has bugs, people get phished, and the defender has to be right every time while the attacker needs one opening. Anyone promising the absence of successful attacks is promising to win an unbounded game forever.
Security vendors know this, which is why "unhackable" appears in marketing and never in threat models. The honest question is not whether someone can get in. It is what they hold when they do.
What does a hacker actually get from sharded storage?
With a normal cloud account, a successful breach yields everything: the complete file, readable or ready to be decrypted later.
With threshold-sharded storage, a file is encrypted on your device, the key is split so that only a minimum number of shares can rebuild it, and the pieces live with independent providers. Breach one node and you hold a fragment that is, in the mathematical sense, nothing: below the threshold, every possible file is equally consistent with the fragments you have. This is called information-theoretic security, and it is the strongest confidentiality guarantee cryptography has. It does not depend on the attacker's budget, their hardware, or the quantum computers of 2040, because the missing information simply is not in what they stole.
That is the claim worth making. Not "you cannot get in," but "getting in stops paying."
How much more effort does the hack actually take?
Count the independent things an attacker must defeat, because independence is what multiplies cost.
Breaching a mainstream drive account is one campaign: one provider, one credential system, one jurisdiction. Reconstructing a threshold-sharded file means compromising several unrelated providers, running different software, in different legal regimes, at the same time, and then also obtaining the owner's signing key, because nodes release shards only against a passkey signature.
Each requirement is a separate campaign against a separate target. Two independent one-in-a-hundred events are a one-in-ten-thousand event. That is the arithmetic difference between a hard target and a pointless one, and it is why sharding is the strongest storage concept available for confidentiality.
What are the honest limits?
Three, and any sharding vendor who does not volunteer them is overselling:
- Reaching the full threshold works. An attacker who compromises enough independent nodes and obtains the signing key can reconstruct. The architecture makes that a coordinated multinational operation instead of a Tuesday, but it does not make it impossible.
- Your device is the last mile. Files are reconstructed and decrypted on your machine. Malware that owns your device at that moment sees what you see.
- Self-custody is real custody. No administrator can reset your access, which also means none can rescue it. Set up multiple authenticators before you need them.
A system described with its limits is a system you can actually evaluate. That is the standard Untrace's own documentation holds itself to.
So what should you actually look for?
Not "unhackable." Look for the property that survives contact with a competent attacker: no single place where the whole file and the whole key exist together. Everything else, the audits, the certifications, the padlock icons, sits on top of whether that property holds.
Encryption alone does not give it to you. Distribution alone does not give it to you. Threshold sharding gives you both halves: the file never whole, the key never whole, and the fragments individually meaningless forever.
Further reading: Why Encryption Is Not Enough, Threshold Thinking for People Who Do Not Like Math, and the FAQ for the direct version of every claim on this page.