Bill Gates stated that the primary risk associated with artificial intelligence currently stems from malicious actors utilizing models without adequate safeguards. He noted that the threat of AI escaping human control remains a long-term concern, whereas the immediate danger lies in the misuse of existing technology.
During an interview on CNN, Gates highlighted that cyberattacks are the most likely area to experience significant negative impacts over the next year. He identified a business model used by attackers that involves either extortion through threats or market manipulation to influence stock prices and betting outcomes.
Industry concerns align with these warnings, as evidenced by an open letter from OpenAI and over 100 organizations in August regarding the rise of sophisticated AI-enabled cyber threats. Research from TeamT5 indicates that state-linked groups have already doubled their attack volume by delegating routine tasks to AI models.
Data from Chainalysis shows a 440% increase in on-chain malware instructions following the release of open-weight models lacking malicious code restrictions in mid-2025. Additionally, CrowdStrike has linked recent breaches at South Korean banks to attackers employing AI tools for financial gain.
