Microsoft co-founder Bill Gates has warned that the immediate danger of artificial intelligence stems from bad actors exploiting models with stripped safeguards, rather than autonomous systems escaping human control. Speaking to Jake Tapper on CNN's State of the Union, Gates emphasized that cyberattacks and market manipulation represent the most critical risks over the next year.
Gates Identifies Cyberattacks and Market Abuse as Immediate Threats
During the interview, Tapper questioned whether global leaders were doing enough to prevent AI-enabled bioterrorism, to which Gates answered that current efforts remain insufficient. Gates explained that while developers attempt to block threats like cyberattacks and bioterrorism using guardrails, those protections can currently be stripped out. "...the real danger right now is not that AI is getting out of control," Gates stated, pointing out that autonomous risks remain years away.
Instead, Gates highlighted two specific business models driving misuse: extortion schemes demanding payment under threat, and attacks designed to move stock markets or affect betting markets. "When you have bad people, no safeguards, and then the latest models, that’s where, over the next year, primarily in the cyber area, we are going to have a lot of negative surprises," he added. These vulnerabilities arrive alongside growing security issues, such as malicious mobile software exploits and broader hardware wallet supply chain attacks.
Security Research Confirms Escalating AI Exploitation
Industry research aligns with Gates' warnings regarding un-safeguarded AI models. In August, OpenAI published an open letter on cyber defense signed by more than 100 organizations, noting that "in the coming months, AI-enabled cyber attacks will become far more widespread and sophisticated..." Furthermore, recent investigations highlight how guardrails are already being bypassed across tech sectors, echoing concerns raised when researchers highlighted issues with unverified AI prompts and outputs.
Key data points tracking this trend include:
- Taiwanese cybersecurity firm TeamT5 reported that Chinese state-linked hacking groups doubled their attack volume after handing routine work to AI tools.
- Chainalysis observed a 440% increase in malware instructions written on-chain following the arrival of open-weight Chinese models with no limits on malicious code in mid-2025.
- Cybersecurity firm CrowdStrike tied several South Korean bank breaches to a likely financially motivated attacker using AI tools.
Why It Matters
Gates' remarks underscore a shift in AI risk assessment from speculative existential threats to immediate operational vulnerabilities. For financial markets and digital asset security, open-weight models lacking safeguards lower the technical barrier for automated exploitation and market manipulation. Institutional investors and blockchain protocols must prepare for an accelerated arms race between AI-assisted defense mechanisms and sophisticated cyber criminals.
