Updated
Updated · Financial Times · Aug 27
AI Cuts Hacking Costs as US Healthcare Breaches Averaged $7.4 Million in 2025
Updated
Updated · Financial Times · Aug 27

AI Cuts Hacking Costs as US Healthcare Breaches Averaged $7.4 Million in 2025

3 articles · Updated · Financial Times · Aug 27

Summary

  • AI-augmented attacks are making cyber intrusions faster and cheaper, while forcing defenders to spend more as models uncover vulnerabilities and generate exploit code in a fraction of the old time and cost.
  • Open-weight models run on private hardware lack the safeguards found in mainstream subscription tools, widening access to advanced attack techniques beyond trained cybersecurity professionals.
  • US healthcare logged 642 ransomware and data-breach incidents reported to the FBI in 2025—more than any other critical-infrastructure sector—with average losses of $7.4 million per breach.
  • Defenses are lagging: the author estimates AI systems remain two to three years away from patching software at the pace needed to match the surge in vulnerabilities.
  • US public-sector resilience has weakened after federal funding for the Multi-State Information Sharing and Analysis Center ended in September 2025 and CISA was cut back, undermining the shared defenses the author says are needed.

Insights

If open-source AI strips away safety guardrails, what stops hackers from automating global ransomware attacks?
With AI discovering vulnerabilities in minutes, are human-led cyber defenses already obsolete?
As AI outpaces traditional security, could the next major hospital blackout be triggered by a semi-autonomous algorithm?