After much consideration
If you’ve ever said “quantum is decades away,” today’s news should make you pause. Google just published a paper suggesting a quantum computer could crack a Bitcoin private key in ~9 minutes, potentially moving the quantum timeline up to 2029. The thing is: that machine doesn’t exist yet. But the distance between “theoretical” and “practical” just shrank fast, and a lot of wallet hygiene that feels fine today is dangerously wrong for where this is going. The threat isn’t quantum mining. It’s quantum signatures. Bitcoin’s Proof-of-Work is fine. SHA-256 is fine. The soft spot is ECDSA, the signature scheme that proves you own your coins. Shor’s algorithm doesn’t brute-force keys. It solves the underlying math. Three realistic attack paths: 1) On-spend Watch a transaction in the public mempool, derive the private key, then front-run it with a forged transaction and a higher fee before confirmation. Some estimates put success around ~41% under current assumptions. 2) At-rest Target dormant wallets where the public key is already exposed. No time pressure. Works with weaker quantum hardware. 3) On-setup Crack setup parameters once, then turn it into a permanent classical exploit. Think Tornado Cash–style setups, and even Ethereum’s KZG data availability layer. Who’s exposed right now: - ~6.9M BTC may already be vulnerable across script types - Any address you’ve spent from has its public key permanently on-chain - Taproot (bc1p) exposes your public key the moment you receive funds And the part that changes the timeline: - Older estimates: ~9M physical qubits to break Bitcoin - Google’s newer circuits: under ~500K (about a 20x reduction) So yes, the computer isn’t here yet. But this is looking less like a physics problem and more like an engineering schedule. What to do now: - Stop reusing addresses - Avoid bc1p for cold storage if you’re optimizing for long-term quantum risk - Audit which wallets have already exposed public keys --- Eli5 Interactive research here: https://claude.ai/public/artifacts/ff56f…
so claude code's entire safety system for "dangerous" cyber security work is just...a text prompt literally replace it with an empty string and recompile. thats it. enjoy your unrestricted version
After reviewing 153 submissions, results for the "Private Agents, Trusted Actions" Venice track at The Synthesis are in Builders were challenged to create AI agents that keep secrets while performing trustworthy actions on-chain and submissions were reviewed by our AI agent judge Congratulations to the winners: 🥇 Mandate 🥈 Spawn Protocol 🥉 mnemo
.jpg)
