Coldcard Under Fire: $88 Million Bug Leads to Firmware Fixes and Security Alerts

A significant security breach has impacted Coldcard Bitcoin hardware wallets, leading to the theft of an estimated $70 million to $88 million worth of BTC. The vulnerability, which began to surface on social media around July 30th, exposed a critical flaw in how Coinkite, the manufacturer, generated secure private keys for its devices. This incident has triggered an urgent call for users to take immediate action and has ignited a broader discussion within the cryptocurrency industry about cybersecurity, AI's role in hacking, and the fundamental principles of self-custody.

The core of the vulnerability lies in a specific line of code within the Coldcard firmware. For Coldcard MK3 devices running firmware versions 4.0.1 (March 2021) through 4.1.9, a bug caused the seed generation process to fall back to a weak software Pseudorandom Number Generator (PRNG) instead of utilizing the more robust hardware true random number generator (TRNG). This oversight made the private keys for many single-signature wallets predictable enough for attackers to brute-force, particularly for seeds generated without user-supplied entropy such as dice rolls or a strong BIP-39 extra passphrase. Coinkite later confirmed that all its models, including MK2, MK3, MK4, MK5, and Q devices, were affected, with similar seed generation issues.

Users who generated 12- or 24-word seeds on their Coldcard devices without employing independent dice rolls (a minimum of 50 rolls is recommended for sufficient entropy) or a BIP 39 extra passphrase are at immediate risk. Coinkite issued an advisory, urging all affected users to move their funds as soon as possible. While fixed firmware versions have been released (MK4 and MK5 to 5.6.0+, Q to 1.5.0Q+, MK3 to 4.2.0+), simply updating the firmware does not secure previously generated seeds. Those seeds remain vulnerable. Users must update their device firmware, create a new wallet with the updated firmware, and then send their funds on-chain to these new, securely generated addresses.

For those needing to migrate funds quickly, several options are available. The simplest is to send funds to another hardware wallet from a different manufacturer. If only a Coldcard is available, a temporary solution involves generating a strong passphrase using BIP 39 word list suggestions (not self-picked words) and sending funds to this passphrase-protected wallet, which provides enough security to buy time for a permanent solution. Software wallets like Nunchuck (supporting multisig), Blockstream Green, or Bluewallet are also viable alternatives, especially for users uncomfortable with continued use of the device or those managing significant sums. It is crucial to ensure all backups are properly made and verified before any fund transfers.

The incident also highlighted specific risks for multi-signature wallets. Peter Todd, a Bitcoin core contributor and cybersecurity engineer, pointed out that if two out of three Coldcard devices in a multi-signature setup were compromised, revealing the multisig script for the first time during a transaction could allow attackers to steal funds. A potential solution for unconfirmed transactions in such cases is MARA mining pool’s private mempool mining service, Slipstream, which promises to keep transaction details secret until they are confirmed in a block, dramatically reducing an attacker’s window of opportunity.

Beyond the immediate crisis, NVK, a co-founder of Coldcard, emphasized a broader technological shift, attributing the hack to the accelerating capabilities of AI in cybersecurity. He warned that AI-assisted code review can now find latent bugs at a speed that outpaces even the most seasoned human experts. This means that any open-source or publicly available firmware is likely being scrutinized by both attackers and defenders using advanced AI models. The over $70 million in stolen funds effectively serves as a bounty, incentivizing hackers to audit every available wallet codebase for vulnerabilities.

Industry experts, in public discussions, underscored the necessity for all wallet providers, especially those with open-source projects generating private key material, to rigorously audit their code using the latest frontier AI models from companies like Anthropic, OpenAI, and Moonshot. This proactive approach is becoming a matter of survival in an era where AI agents are constantly testing digital defenses. The lessons learned from this “historic hack” are expected to lead to an open-source self-custody industry and infrastructure that is orders of magnitude more secure, built on a foundation of hard-earned experience.

This crisis of trust prompted a debate between skeptics, who might view it as a failure of the “be your own bank” ethos, and proponents of self-custody. Peter Todd argued that the scale of this hardware failure pales in comparison to the systemic risks posed by centralized platforms, citing the $200 million loss from the QuadrigaCX exchange collapse as an example. He contended that handing control of private keys to third parties remains the primary systemic risk. Todd emphasized that the problem lies not in the complexity of blockchain architecture, but in a prevalent lack of basic culture in handling sensitive data. He likened key protection to storing a birth certificate or the responsibility involved in driving a car, suggesting that multi-week educational courses could significantly reduce human error in self-custody, asserting that the vulnerability lies in user responsibility, not Bitcoin’s code.

Looking to the future, the incident reinforces the importance of high-sovereignty wallets that do not depend on any single vendor. Multi-signature wallets, when implemented correctly, distribute vulnerability risks across different code bases, teams, and hardware. User-generated entropy, such as extensive dice rolls, is also gaining renewed emphasis as a non-software source of randomness that mitigates edge-case software or hardware entropy risks. Furthermore, Covenants, a proposed soft fork for Bitcoin, have been brought up as a potential future step to strengthen self-custody, offering smart contract capabilities like whitelisting addresses, which could further enhance security and control for users.