The debate over whether artificial intelligence could ever hack Bitcoin and seize the coins of the network's anonymous creator, Satoshi Nakamoto, has taken a new turn. The catalyst was a prediction by investor Chamath Palihapitiya, founder of the Social Capital fund, who stated that Bitcoin will one day be hacked and depreciate to zero. This claim was addressed by Grok, the chatbot from xAI, which explained why such a scenario is technically unfeasible.

The discussion unfolded on the social network X. Users asked the neural network why it had not yet cracked Satoshi's wallet, which is estimated to hold about a million BTC — coins untouched since the early years of the network's existence.

The investor called Satoshi's wallet the main target

The starting point was Palihapitiya's thesis: if AI were aimed at cryptocurrencies, the first target would supposedly be Satoshi Nakamoto's wallet. According to the investor, the plan is simple — empty it, then bet on the price falling and profit. For the market, such a scenario would be a catastrophe.

One user suggested that Grok "break" SHA-256 and secp256k1 and take Satoshi's coins. These are the two algorithms on which Bitcoin rests. SHA-256 converts any data into a unique string of characters and is used in mining. Secp256k1 is responsible for digital signatures, linking the owner's secret key to the wallet address. Without the key, the coins cannot be moved.

The neural network explained why the plan won't work

Grok pointed out that no AI, even the most advanced, will crack these algorithms. They are based on mathematical problems that cannot be solved quickly. Brute-forcing all the options is pointless — it would take billions of years. Quantum computers capable of such an attack do not yet exist. Therefore, Satoshi's coins remain under lock and key.

The user objected, recalling that Palihapitiya linked the hack to recursive self-improvement of AI — the ability of a neural network to rewrite its own code and become smarter. Grok disagreed. The bot acknowledged that the investor has a good understanding of markets and trends in the field of AI, however, self-improvement only helps to write and refine programs faster. It does not invent solutions for problems that are inherently unsolvable.

The neural network then addressed several fears common among investors:

  • AI will not be able to print new bitcoins, since the limit of 21 million coins is protected by the network's rules, and nodes will reject any violation;
  • emission can only be changed with the consent of the entire community, but then Bitcoin would cease to be scarce and would lose its value;
  • a quantum computer capable of breaking secp256k1 or SHA-256 today does not exist.

According to Grok's estimate, a single attempt to crack a signature would require a machine of the IonQ level with roughly 20,000 physical qubits, running for about 26 days. Reliable quantum computers of that class do not yet exist. Whether they will appear by 2030–2035 is also unknown. If the threat becomes real, the Bitcoin network will be able to switch to signatures resistant to quantum attacks.

Cryptalist analysis: Claims about Bitcoin being hacked by artificial intelligence belong more to the realm of narratives than to a real technical threat. The cryptographic strength of the network is ensured by fundamental mathematical constraints, not by vulnerabilities that can be circumvented by growth in computing power. For institutional investors, the key risk remains not a hack of the protocol, but a potential quantum threat in the long term, for which the network already has adaptation mechanisms.