Compiled & Organized: Deep Tide TechFlow

Host: Austin Campbell (host, Zero Knowledge Group), Ram Ahluwalia (CEO of Lumida Wealth), Chris Perkins (Franklin Crypto)
Guest: Alex Thorn, current Managing Director & Head of Firmwide Research at Galaxy Digital, previously worked at Fidelity Investments for 12 years, known as the “Bitcoin Viking”
Core Topics: Coldcard hardware wallet vulnerabilities, four waves of attack chain tracing, AI safety rails backlash against defenders, the boundary between self-custody and institutional custody, Federal Reserve interest rate environment
Program Name: Unchained · Bits + Bips · The Coldcard Hack with Galaxy’s Alex Thorn
Conflict of Interest Statement: Alex Thorn is employed by Galaxy Digital, which provides institutional-level cryptocurrency custody, trading, asset management, and ETF-related services; Chris Perkins works at Franklin Crypto under Franklin Templeton, participating in crypto ETF and other businesses; Ram Ahluwalia is the CEO of Lumida Wealth, providing cryptocurrency asset management services. This discussion involves a comparison between self-custody and institutional custody, and the guest’s views are related to the business interests of their institutions.
Summary of Key Points
- Coldcard introduced a new random number generator (RNG) in the firmware update on March 17, 2021, but wiring errors caused it to never be truly called, and the system silently reverted to a weak pseudo-random number generator.
- Attackers exploited this vulnerability to bulk compute possible private and public keys, subsequently emptying affected single-signature (single-sig) addresses. The issue lies not in the Bitcoin protocol itself, but in the underlying cryptographic implementation.
- Alex Thorn estimated on the program that the first three waves of attacks caused approximately 1400 BTC in losses; if the still-confirming fourth wave is included, the total scale approaches 2000 BTC, equivalent to over 100 million dollars, marking an unprecedented scale for vulnerabilities in distributed self-custody hardware wallets.
- The majority of victims are long-term holders, with an average coin age of nearly four years. They did not click phishing links nor did they disclose mnemonic phrases; they simply trusted the random number generation of the hardware wallet.
- Because thousands of unrelated cold storage addresses were stolen, on-chain tracking is much more difficult than for stolen DeFi bridges. Alex has submitted approximately 600 suspicious addresses to US federal law enforcement, exchanges, and on-chain analytics firms such as Chainalysis, TRM Labs, and Elliptic.
- Cutting-edge US AI models refuse to assist in tracking and auditing code due to safety rails, forcing red teams to use Chinese open-source models like Kimi and GLM for defense analysis.
- The incident is driving Bitcoin away from self-custody towards exchanges and ETFs. Alex believes that multi-signature collaborative custody (such as with Casa, Unchained, etc.) remains a secure self-custody solution, but for ordinary people, regulated intermediaries like Coinbase, Bitkey, or ETFs might be more realistic options.
- Host Chris Perkins emphasized that the only free lunch in finance is diversification. Do not put all your Bitcoins in a single wallet or a single ETF.
Highlights of Opinions
“If you are still keeping coins in a Coldcard single-sig address, please transfer them as soon as possible.”
“These people did nothing wrong. They did everything right. They are not DeFi yield farmers or meme coin speculators; they are just hardworking individuals saving Bitcoin and putting it into cold wallets.”
“This is not a Bitcoin problem, it’s a cryptography problem. If email service providers used the same random number generator, your emails would be cracked too.”
“The cutting-edge AI models in the US are unwilling to help us track these stolen funds, and our red team can only turn to Chinese open-source models. This is absurd.”
“I will not leave the self-custody train. But if your mother or father asks you how to store Bitcoin, I might suggest they open a Coinbase account.”
“The hedge fund industry is the most glamorous yet the most painful. You think you are the master of the universe, yet you wake up at three in the morning to check Korean stock prices and at six, you check what’s happening in the world.”
Body
1. Market Snapshot: BTC Holds at $63,000, AI Fund Situational Awareness Loses $35 Billion in One Month
At the beginning of the program, Austin Campbell presented a market snapshot: Bitcoin fluctuated around $63,000 on that day, Ethereum around $1850, and the 10-year US Treasury yield remained in the relatively high range of 4.72% to 4.75%, with the market pricing in over 50% hawkish expectations from the Federal Reserve.
The top story of the day was the collapse of the Situational Awareness fund. The fund’s size dropped from about $45 billion to approximately $10 billion, with a monthly AUM evaporation of about $35 billion, a decline of 67%. The prime broker's margin calls forced the fund to sell its listed stock portfolio at distressed prices to Ken Griffin's Citadel, with remaining assets mainly consisting of Anthropic's private equity and private placement instruments.
Ram Ahluwalia believes this was not Citadel’s conspiracy. “Micron’s stock price dropped double digits after earnings reports, just like the previous quarter. Momentum trades peaked at three standard deviations and then collapsed.” He compared Leopold to the previously collapsed Asymmetric fund (long Solana, short ETH), saying “it’s essentially the same leveraged bet.” Ram also noted that Leopold is only 24 years old, and his fiancée is the Chief of Staff to the CEO of Anthropic, “being too close to this circle prevents them from seeing outside of the AI bubble.” He estimates that many new funds were invested at high points in the past three months, and the liquid book was taken by Citadel at distressed prices, “these people’s principal is most likely gone.”
Alex Thorn was asked about the fundamentals of AI infrastructure and data centers. He assessed this collapse as “80% technical, 20% fundamental”: “During the rise, Situational Awareness’s own positions were probably one of the main driving forces.” But he emphasized that the four major cloud providers are still guiding substantial capital expenditure growth, and the cybersecurity and computing demands related to AI are far from being met. “This is somewhat similar to the dot-com bubble and also a bit like Bitcoin in 2017. Many people said that was the top, but today Bitcoin is more than three times its 2017 peak.”
Chris Perkins added from a risk management perspective: “Young markets with leverage will amplify volatility in both directions. The directional judgment may be correct, but if the positions and risk control are wrong, the outcome will still be zero.” He also mentioned that Citadel, Jane Street, and Millennium are now playing the role of “last lender,” “the era when regulators and investment banks gathered everyone to the New York Fed for money like in the LTCM days is gone. Now, just a phone call to Citadel will do.”
2. Coldcard Vulnerability: A Miswired Random Number Generator Turns Single-sig Addresses into Cash Machines
Before discussing Coldcard, Alex Thorn offered an urgent piece of advice: “If you are still keeping coins in a Coldcard single-sig address, please transfer them as soon as possible.”
Austin Campbell summarized the incident with a sentence: Coldcard has always been regarded as the “gold standard” among Bitcoin hardware wallets but has caused the Bitcoins of Mark III, Mark IV, and Mark V users to be emptied in bulk due to entropic vulnerabilities, with affected addresses exceeding one thousand and the loss figures continuing to rise.
Alex explained that generating cryptographic keys requires sufficient entropy from a random number generator. The weaker the entropy, the easier it is for keys to be derived through brute force. Coldcard added its own random number generator in a firmware update on March 17, 2021, but a wiring error caused it to never be properly called. The system would silently fail and revert to a pseudo-random number generator with extremely weak entropy.
“Once the attackers realized this, they only needed computing power to bulk generate candidate private keys and public keys, then scan which addresses in the entire derivation path had balances. Once hit, they would initiate transactions to transfer the coins away.”
Alex emphasized that the cruelty of this type of attack lies in the fact that the victims did not make any mistakes: “They did not disclose their mnemonic phrases, did not click phishing links, and did not share private keys in groups. They were just hardworking individuals making an effort to save Bitcoin and put it into cold wallets. These people did nothing wrong; they did everything right.” He pointed out that the average dormancy time of the swept coins is nearly four years, indicating that most victims are long-term holders.
Chris Perkins added that this is essentially a problem of cryptographic implementation, not an issue with Bitcoin, Ethereum, or any blockchain protocol. “If an email service provider used the same random number generator, your encrypted emails would also be cracked.”
Alex also mentioned that this vulnerability went undetected for five years, possibly related to codebase governance: “It’s not fully open-source, and there are many uncommented commits on the main branch.” He acknowledged that AI may have accelerated the discovery of the vulnerability, but “without AI, it could still be found; the problem is there aren’t enough eyes reviewing this code.”
3. Four Waves of Attacks and On-chain Tracking: The “Silent Theft” of Thousands of Cold Wallet Addresses
Alex revealed that he had been conducting on-chain tracking over the past few days, identifying victim and attacker addresses and submitting these addresses to US federal law enforcement, major exchanges, and analytics firms like Chainalysis, TRM Labs, and Elliptic, “so that once these coins enter centralized intermediaries, there is still a chance for recovery.”
He divides the currently known attacks into four waves, but the fourth wave has not been fully confirmed:
- First and Second Waves: Large amounts of victims' coins were aggregated to one or two collector addresses, then transferred to one or two attacker-held addresses with a similar structure. Alex suspects this may have been done by the same group. These two waves account for over 70% of the total loss amount.
- Third Wave: The structure is completely different. Attackers established a separate vault for each victim, “if your 10 addresses are swept, they will first aggregate to an intermediate address that belongs only to you, then enter a vault that belongs only to you.” Alex estimates that the third wave involved about 293 victims and 293 vaults.
- Fourth Wave: Alex identified it solely through burst patterns, with currently 208 transactions sweeping approximately 389 BTC from 462 suspected victim addresses in about 2.5 hours, but no victims have come forward to confirm yet.
In addition to these four waves, Alex discovered at least 14 other identifiable attack patterns, “I have temporarily named them footprint A to L.” Victims within these patterns have already filed police reports, but individual scales are small, and a total volume has not yet been extrapolated.
Alex provided the latest estimate on the program: as of the end of the third wave, losses are close to 1400 BTC; if the fourth wave is confirmed, the total scale may approach 2000 BTC; even without counting the fourth wave, it can be placed around 1600 BTC. “For vulnerabilities in distributed self-custody storage, this is the largest scale I have ever seen.”
The difficulty of on-chain tracking lies in the fact that these addresses were originally unrelated. “If it were a DeFi bridge being hacked, we would just need to monitor the smart contract address to see where the funds flow. But here, every address is a user's cold storage address, and we must rely on the attacker’s transaction fingerprints, victims' reports, and fee overpaying behavior patterns to reverse engineer.”
4. AI Safety Rails Backlash: US Models Refuse to Assist, Defenders Can Only Turn to Chinese Open-source Models
Austin Campbell shifted the topic to AI and security: “In this battle of offense and defense, does AI help attackers or defenders more?”
Alex Thorn’s answer was surprising: defenders are instead blocked by the safety rails of US AI models. “The red team that is auditing all hardware wallet codebases has been forced to use Chinese open-source models like Kimi and GLM because cutting-edge US models do not allow for this type of cybersecurity analysis at all.”
He gave an example using his toolchain: “I have a complete Bitcoin analysis system on my local machine, a 3.7 billion line, 1.7 TB Postgres database that records every transaction and state change in Bitcoin’s history. I asked Claude Fable 5 to help me query some local data, and half the time I receive a rejection due to security cards. This is data on my own machine; I haven't even browsed any webpages.”
Chris Perkins believes that policy needs to redefine the boundaries between “defenders” and “attackers.” “We should allow white hats to take proactive actions, just as the Constitution allowed privateers long ago. This is no different from a burglar entering your home or your car being stolen; it cannot be treated as normal loss just because it happens in the crypto world.”
Ram Ahluwalia provided a broader technological historical perspective: “Most technologies favor the attackers. Stuxnet, airplanes, cyber weapons—history has repeatedly proven that defenses are too broad, and attackers only need to find one breach.” He pointed out that the US still lacks a competitive open-source model, “this is a public goods problem, like parks; everyone benefits, but no one is incentivized to pay to maintain it individually. China, needing to catch up on hardware and original research and development, has taken a different approach. I think the Department of Defense and DARPA should seriously consider funding an open-weight model.”
Austin mentioned that the founder of Hugging Face said in a political program, “AI will be the biggest asset in DeFi in the long run; we just need to get over this hump first.” Alex agreed with this assessment: “Once these security issues are resolved, they are resolved. AI will ultimately be very helpful for security.”
5. Self-Custody vs Institutional Custody: How Should Ordinary People Choose?
Austin Campbell admitted that he currently does not have a standard answer: “If my mom, dad, or friends ask me how to store Bitcoin, I can't answer.”
Alex Thorn believes that most losses in self-custody funds actually come from phishing attacks or user errors. “It used to be said, would you let your grandmother use a hardware wallet or let her open a Coinbase account? I genuinely feel that for most ordinary people, Coinbase or Bitkey is more suitable.” He specifically mentioned that Bitkey is the first hardware wallet he is willing to recommend to older family members.
However, he emphasized that Bitcoin's multi-signature collaborative custody solutions (Casa, Unchained, Unchuck, Leona Wallet, AnkerWatch, etc.) are still very secure, “it distributes private keys across multiple hardware wallets and custodians, preventing single point failures.” He said he would not leave the self-custody train, “I’m like Jordan Belfort shouting, we’re not effing leaving.”
However, the reality is that events have already sparked a flow of funds back from self-custody to exchanges and ETFs. Alex quoted data from Timechain Index and CryptoQuant: since last Friday, about 22,000 Bitcoins have flowed back to exchanges, with small transactions (1 BTC) into exchanges reaching a multi-year high. “This is a reversal of flow after FTX. People are fleeing self-custody and flocking to exchanges.”
Alex has a tolerant view of this phenomenon: “Some Bitcoin maximalists may look down upon people using ETFs, thinking they are not true Bitcoiners. But that's their money. If you lack the education and training required for self-custody, that’s okay.” He himself holds Bitcoin through multiple means including River, Fidelity Digital Assets, ETFs, and more.
Chris Perkins took over, emphasizing that “the only free lunch in finance is diversification.” He cited the news about NBA star Giannis Antetokounmpo diversifying cash across 50 banks, saying, “Greece has a unique memory of banking crises, and the Cyprus haircut was also one of the reasons the Winklevoss brothers initially favored Bitcoin. Do not put all your Bitcoins in one place, whether in a self-custody wallet or an ETF.”
Ram Ahluwalia added from the perspective of institutional investors: “The core demand for institutions is having someone to sue. They do not need to control the private keys themselves; they need a capitalized counterpart that can be sued.” Alex also mentioned that he has consistently recommended qualified institutional custodians to high-net-worth clients, family offices, and funds, “now most of the major Bitcoin ETFs in the US adopt multiple custodians for the same diversification logic.”
6. Federal Reserve and Interest Rates: An Environment Without “Fed Put”
At the end of the program, Austin Campbell shifted the topic to the Federal Reserve. The FOMC maintained interest rates in the range of 3.50% to 3.75%, with a voting result of 9 to 3, with officials like Kashkari opposing. Kashkari publicly stated that investment in data centers adds new demand-side pressure to inflation, and he would prefer a slight tightening before receiving more data.
Alex Thorn believes that the Federal Reserve is allowing the market to do part of its work. “There is no longer a Fed put; this is the market consensus.” He refused to predict the Bitcoin price for this year at the end of last year, saying, “At that time, I said the investment environment was very uncertain, and now it seems that judgment was correct.”
He also mentioned that the US Treasury seems to want to lower interest rates, “power is shifting from the Federal Reserve to the Treasury.” At the same time, the yen carry trades have not fully unwound, “Scott Bessent buying yen and selling euros, Trump saying we love the Japanese, except for Pearl Harbor. In this environment, the direction of interest rates is very difficult to judge.”
Ram Ahluwalia believes that Kevin Warsh's criticisms regarding the Federal Reserve's accountability have some merit: “US inflation has been above 2% for five consecutive years; that's a fact. But long-end yields are rising spontaneously, and the Fed's forward guidance may be replaced by market-formed reaction functions.”
Chris Perkins concluded: “We need better policy response. The thefts in the cryptocurrency space are still thefts; they cannot be tolerated just because they occur on-chain.”
免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到support@aicoin.com,本平台相关工作人员将会进行核查。