The North Korean hacking group Kimsuky is now using artificial intelligence to design cyberattacks. They have built three local AI environments - Ollama, GPT4All, and Msty - that run without needing to send queries to external providers. With these local models, they can write malware code, analyze data, and automate phishing campaigns without being detected by monitoring systems or facing account suspensions from commercial AI services.
Responding to this new tactic, the Bitcoin Policy Institute led more than 40 digital asset organizations in sending an open letter to leading AI laboratories on August 10, 2026. The list of signatories includes major names such as Block, Coinbase, Strategy, MARA, Galaxy, BitGo, and hardware wallet maker Trezor. They are urging that verified open-source security researchers be granted controlled access to frontier AI models.
The coalition is not asking for unrestricted public access. Their specific requests include early access to frontier cybersecurity models, guarantees of adequate computing capacity, secure research environments, and direct communication channels with security teams at AI laboratories. As of the time of writing, no AI labs have announced an official program in response to the coalition’s demands.
Not Just a Theoretical Threat
Attacks on crypto projects this year have been costly. Data from Immunefi recorded losses of approximately $110 million from various hacks in July 2026 alone. As of August 3, 164 security incidents have occurred. This trend has the potential to set a record, with theft incidents exceeding $1 million projected to reach 114 cases throughout the year.
Internal research shows how quickly AI can dissect code vulnerabilities. In AI-assisted research lasting less than 30 hours, the Bitcoin Red Team detected 4,962 potential issues across 390 Bitcoin projects. Of those, 720 findings were classified as high or critical severity.
Hardware-level vulnerabilities also add to the long list of weaknesses. A report from Galaxy Research states that Coldcard wallet firmware has harbored an entropy bug since 2021, which weakens the seed phrase generation process. Consequently, confirmed thefts from this vulnerability have reached 1,596 BTC, with additional waves of attacks potentially increasing losses by up to 2,055 BTC.
New Defense Capital
To keep pace with the hackers’ speed, industry players have begun injecting self-funded research capital. A Bitcoin security consortium - consisting of Strategy, BlackRock, Coinbase, and six other companies - has pledged to provide $15 million over the next three years specifically for long-term security research.
Outside of that consortium, Galaxy has opened a separate $5 million Bitcoin security fund. The allocation focus of this fund includes the development of new signature systems, the creation of wallet migration tools, conducting routine audits, and research into quantum computing resistance.
The arms race between hackers and the crypto ecosystem has now shifted to the realm of artificial intelligence. If third-party security researchers continue to fall behind state actors who operate freely without commercial service restrictions, the ecosystem’s defense layers will face a severe test.
Sourced from crypto.news.
Disclaimer: This article is for informational and educational purposes only, not financial advice. Cryptocurrency assets are highly volatile and carry significant risk. Always do your own research (DYOR) and never invest more than you can afford to lose.
