Overview
- Germany’s federal cyber agency, the BSI, reported this week that current AI systems can quickly find software vulnerabilities, write matching exploit code and scale attacks far faster than before.
- BSI leaders and senior security officials told a national conference that the time from vulnerability discovery to exploit has shrunk and that defenders must assume many patched zero‑days were already abused.
- Officials urged Europe to fund homegrown large models, with BSI executives estimating the task will need investment on the order of hundreds of billions of euros to close the gap with the United States and China.
- Reporting shows frontline governance and training gaps: childcare workers routinely use free AI tools like ChatGPT to draft reports, exposing thousands of children’s personal data without clear safeguards.
- Business adoption is surging—AI use in firms roughly doubled in a year—but research and case studies show agentic systems struggle with long‑horizon management tasks, so urgent defensive fixes and staff training are immediate priorities.