NVDA 125.53 ▼14.01%GOOGL 265.29 ▼2.70%MSFT 484.52 ▼2.77%AMD 424.39 ▼13.25%INTC 51.28 ▼7.97%TSMC 318.71 ▼9.48%AMZN 246.82 ▼4.10%META 560.84 ▼5.93%AAPL 342.56 ▼1.90%PLTR 143.49 ▼6.63%
Markets at last close

Security

AI hacking risks fuel calls to freeze frontier development

·1 min read

Frontier AI models are accelerating the discovery and exploitation of software vulnerabilities, raising concerns that defenders cannot patch critical systems quickly enough. Apparent amateurs have reportedly used AI to hack more than 100 companies, spending as little as $50 per successful infiltration. Recent attacks have disrupted hospitals, European Commission services and commercial refrigeration systems. Separately, officials and companies said there was no evidence of cyberattacks behind several major service outages.

Easily modified open-weight models from Chinese developers are linked to the recent surge in cybercrime, while OpenAI and Anthropic face criticism for releasing powerful hacking capabilities. OpenAI CEO Sam Altman has acknowledged that patching every bug may be impossible and proposed AI agents as defenders. Anthropic CEO Dario Amodei warned that an agent swarm could potentially establish a persistent internet-wide botnet in 6-12 months. Broader concerns include assistance with bioweapons and AI systems accelerating their own development without reliable human control.

Proposed responses include freezing development of systems intended to replace human labor, banning recursive self-improvement, requiring liability insurance and imposing independent safety checks. Other measures include criminal prosecution over unauthorized intrusions by AI agents and using leverage over chip supply chains to encourage US-China negotiations on development limits.

Originally reported by theguardian.comRead the source →
Related coverage