we have a year to fix security everywhere
Summary
The article argues that frontier LLMs, including open-weight models like GLM 5.3-flash, create immediate cybersecurity risks by enabling powerful, low-barrier hacking capabilities. It advocates for urgent policy, governance, and deployment strategies—from government incentives to industry patching and supply-chain security—while emphasizing that deployment, not just discovery, is the hard part. The piece urges proactive, structured defenses and ongoing security posture improvements to stay ahead of rapidly advancing AI threats.