Coldcard RNG Flaw Drains $130M in Bitcoin

A Five-Year Firmware Disaster Exposes Bitcoin's Security Blind Spot 🚨

In July 2026, something unprecedented happened in the hardware wallet space. Attackers drained over 1,082 Bitcoin from nearly 1,200 devices in just 41 minutes—without stealing a single device, guessing a PIN, or touching the Bitcoin protocol itself. They did it from a laptop using an offline seed-reconstruction tool, exploiting a firmware error that had been silently compromising Coldcard wallets since March 2021. The total confirmed theft now stands at 1,596 BTC from approximately 7,300 addresses, with suspected losses potentially reaching 2,055 BTC—close to $130 million at current valuations.

This wasn't a sophisticated zero-day exploit or a breakthrough in cryptanalysis. It was something far more troubling: a single misplaced character in the firmware code that nobody caught for five years.

The Macro Check That Broke Everything 💻

The vulnerability originated during a seemingly routine firmware update. Coinkite, the Canadian manufacturer behind Coldcard, was integrating a new cryptographic library called libngu into their codebase. During this migration, a critical mistake slipped through: an incorrect conditional check caused the seed-generation function to bind to a software-based pseudo-random number generator called Yasmarang instead of Coldcard's hardware true random number generator.

Here's where it gets interesting. The production configuration for affected Coldcard models sets a macro called MICROPY_HW_ENABLE_RNG to zero, because Coldcard uses its own separate hardware random number generator wrapper. The libngu library checks for this macro using a #ifndef conditional—which only tests whether the macro is defined, not whether its value is zero or non-zero.

Because the macro was defined with a value of zero rather than absent entirely, libngu treated the hardware source as available and silently redirected seed generation to MicroPython's deterministic Yasmarang generator. This generator produces output seeded only from the microcontroller's unique identifier and timer registers at boot—neither of which provides cryptographic randomness.

The kicker? The hardware RNG continued running perfectly in every other firmware function throughout the entire affected period. Internal code reviews at Coinkite could see the hardware generator was present and functional, never detecting that wallet seed creation had quietly switched to the weaker fallback.

The Entropy Disaster: How Bad Was It? 📊

The impact varied by device model, but all were catastrophically compromised:

• Mk2 and Mk3 devices (firmware 4.0.0 through 4.1.9): Seeds contained approximately 40 bits of effective entropy with zero cryptographic input added to the random number generator output
• Mk4, Mk5, and Q devices: Received minimal contribution from a secure element at boot, resulting in approximately 72 bits of effective entropy instead of the intended 128 bits

To put this in perspective, a properly generated seed phrase draws from 128 bits of uniform randomness, making it computationally infeasible to guess. A seed with only 40 bits of entropy occupies roughly one trillion possible values—a number that modern computers can brute-force in hours rather than millennia.

This represented an exposure roughly 72 quadrillion times weaker than the intended design.

How Attackers Operated Without Physical Access 🎯

The exploitation method was remarkably straightforward, which makes the incident even more alarming. Attackers didn't need to physically access devices or guess PINs. Instead, they used offline seed-reconstruction tools that could systematically work through the reduced entropy space.

At least 15 distinct attackers have exploited this same flaw independently, suggesting the vulnerability became public knowledge relatively quickly once discovered. The coordinated sweep on July 30, 2026, removed over 1,082 Bitcoin from 1,196 hardware wallets in approximately 41 minutes—a level of sophistication and coordination that indicates professional criminal operations.

Perhaps most concerning: roughly 90% of stolen funds remain unmoved on the blockchain, giving investigators a narrow window to coordinate with exchanges and law enforcement before laundering activities accelerate.

The Industry's Broader Entropy Problem 🔍

What started as a Coldcard-specific incident has evolved into a systemic wake-up call for the entire Bitcoin ecosystem. Block's Bitcoin engineering and security team independently confirmed the flaw and published technical disclosures. But parallel to the Coldcard investigation, a wave of AI-assisted security analysis has been sweeping through other Bitcoin wallet codebases.

The results are sobering: researchers have identified 85+ critical bugs related to entropy handling across the broader Bitcoin wallet landscape—vulnerabilities that conventional review processes had completely missed.

Kraken's chief security officer has called for mandatory independent entropy testing across all hardware wallet manufacturers. This represents a significant shift in how the industry approaches security validation. Previously, most manufacturers relied on internal code reviews and limited third-party audits. The Coldcard incident demonstrates that this approach is insufficient.

What This Means for Bitcoin Security Standards 📋

The Coldcard disaster exposes several critical gaps in how hardware wallet manufacturers approach security:

1. Single Points of Failure: A single misplaced character in a conditional check compromised millions of dollars in Bitcoin
2. Review Process Gaps: Internal code reviews failed to catch the vulnerability despite the hardware RNG being visible and functional elsewhere in the code
3. Lack of Standardization: No industry-wide standard for entropy testing allowed similar vulnerabilities to hide in other wallets
4. Entropy Complexity: The distinction between checking if a macro is defined versus checking its value is subtle enough to evade human reviewers

The Response and Recovery Path 🛠️

Coinkite has released corrected firmware and destroyed all affected-device inventory. However, this creates a critical problem: existing vulnerable seeds require complete wallet migration regardless of which firmware version is currently installed. Users cannot simply update their firmware and assume their Bitcoin is safe—they must move their funds to new wallets generated with corrected firmware.

For users who haven't yet discovered they were affected, the window is closing. As stolen funds begin moving and laundering activities accelerate, the ability to trace and recover Bitcoin diminishes rapidly.

AI-Powered Security Audits: A New Frontier 🤖

The discovery of 85+ additional critical bugs through AI-assisted analysis marks a turning point in cryptocurrency security. Machine learning models can analyze vast codebases, identify patterns associated with entropy vulnerabilities, and flag potential issues that human reviewers might overlook.

However, this also raises questions about why these vulnerabilities weren't caught earlier. The Bitcoin wallet industry has been operating under the assumption that conventional security practices—internal reviews, limited third-party audits, and community scrutiny—were sufficient. The Coldcard incident and subsequent AI findings suggest otherwise.

What Users Should Do Now ⚠️

If you're a Coldcard user, immediate action is necessary:

1. Check your firmware version against the affected range (4.0.0 through 4.1.9 for Mk2/Mk3)
2. Generate a new seed phrase using updated, corrected firmware
3. Migrate all funds from potentially compromised addresses to new wallets
4. Monitor your old addresses on the blockchain for any unauthorized movement
5. Report suspected theft to law enforcement and your exchange

For users of other hardware wallets, this incident should prompt a thorough security review of your chosen manufacturer's entropy handling practices and audit history.

The Bigger Picture: Rethinking Hardware Wallet Security 🌐

The Coldcard incident represents more than just a manufacturing failure—it's a fundamental challenge to how the cryptocurrency industry validates security. Hardware wallets are meant to be the gold standard for Bitcoin security, isolated from internet-connected devices and protected by specialized cryptographic hardware.

Yet a single firmware error managed to compromise the security of hundreds of thousands of users for five years. This suggests that the industry needs:

• Mandatory independent entropy audits before devices reach market
• Ongoing AI-powered security analysis of wallet codebases
• Standardized testing frameworks for cryptographic randomness
• Faster vulnerability disclosure and patching processes
• Better communication between manufacturers and security researchers

Looking Forward: Lessons for Bitcoin Security 🔮

As the Bitcoin ecosystem matures and adoption accelerates, security practices must evolve accordingly. The Coldcard incident demonstrates that even well-established manufacturers with good reputations can make catastrophic mistakes that go undetected for years.

The emergence of AI-assisted security analysis offers hope that future vulnerabilities might be caught before they're exploited at scale. However, this must be paired with industry-wide commitment to rigorous testing standards, transparent security practices, and rapid response protocols.

For Bitcoin users, the lesson is clear: hardware wallets remain essential for security, but they're not infallible. Diversification, regular security audits, and staying informed about emerging vulnerabilities are more important than ever. The $130 million in stolen Bitcoin serves as an expensive reminder that in cryptocurrency, security is not a feature—it's a prerequisite.