AI's cybersecurity capabilities threaten internet fragmentation and raise questions about private control
Source: https://www.theguardian.com/profile/editorial. "The Guardian view on Anthropic’s Claude Mythos: when AI finds every flaw, who controls the internet? | Editorial | The Guardian." April 23, 2026. www.theguardian.com
The Gist
AI can now hack into computer systems automatically, making cyber-attacks much easier for anyone to carry out. This creates a big problem because private companies control this powerful technology, and countries might start building separate, competing internet security systems instead of working together on one shared internet.
Conclusion
The development of AI models capable of autonomous cyber-attacks creates systemic risks that could fragment the global internet and raises serious questions about whether private companies should control such powerful capabilities
Premises
- Anthropic's Claude Mythos can autonomously find zero-day vulnerabilities and exploit them to take over major systems
- AI democratizes hacking by making system breaches accessible to non-experts, turning latent weaknesses into systemic risks
- The US government's selective partnership with Anthropic (sharing only with American organizations and Britain) demonstrates how AI capabilities create geopolitical advantages
- Private companies controlling critical cybersecurity infrastructure poses risks, especially if less responsible actors gain similar capabilities
- Without international coordination frameworks, countries will create separate, competing security systems rather than maintaining a shared global internet
Assumptions
- A unified global internet is preferable to fragmented national security systems
- Private companies cannot be fully trusted with capabilities that affect global cybersecurity
- AI capabilities will continue to spread to other actors beyond current leaders
- International cooperation on cybersecurity is both possible and necessary