North Korea no longer cracks vaults - it befriends the guards. Ari Redbord of TRM Labs explained on Bankless that North Korean operatives now recruit Western proxies to attend conferences, join developer communities, and even make million-dollar investments in target projects.

These proxies operate as sleeper cells, gaining trust over months to access validator keys or administrative privileges. The recent $285 million hack of the Solana-based Drift Protocol occurred during a planned security migration, executed in 12 minutes. Redbord warns these actors are likely embedded in dozens of other teams.

For North Korea, crypto theft is the economy. With virtually no exports, the state has professionalized hacking, selecting agents from childhood for STEM prowess. Over the last five years, these groups have stolen roughly $6 billion to fund weapons and regime survival. They prioritize speed over stealth, converting stolen Ethereum to Bitcoin within 72 hours via bridges like Thorchain before laundering through Chinese OTC brokers.

Roman Yampolskiy dismisses the industry's safety frameworks as futile. He argues no containment mechanism can scale to constrain an agent that outthinks humanity in every domain, putting the probability of catastrophic outcome at nearly 100%. Corporate developers chasing trillion-dollar incentives are merely 'safety washing' their products with surface filters that don't alter underlying goals.

Zico Kolter, who chairs OpenAI's Safety and Security Committee, provides a narrower technical critique. He notes that while scaling compute solves performance issues, it rarely fixes security vulnerabilities. Robustness requires explicit training and architectural guardrails - separate engineering work not emergent from model size. His committee acts as a release brake, reviewing red-teaming reports and possessing the authority to stall a launch if thresholds aren't met.

The attack surface expands with agentic systems. Kolter highlights prompt injection, where malicious instructions embedded in third-party data like emails can hijack an agent to leak API keys or financial data. Permissions become the last line of defense, requiring developers to treat agents as unprivileged users. This hybrid challenge blends AI safety training with traditional cybersecurity.