Tarah Wheeler, information security researcher and chief security officer at TPO Group, warns that AI is dramatically lowering the cost and time required to launch cyber attacks. After a talk to Indiana public officials, an IT director faced pressure to adopt AI while still struggling with basic cyber hygiene like personal device usage. Wheeler notes that AI models can generate exploits far cheaper and faster than human hackers, creating a two‑tier capability gap.

Open‑weight models on personal hardware lack safeguards, putting powerful attack tools in untrained hands. In 2025, U.S. healthcare suffered 642 ransomware and breach incidents—averaging $7.4 million per breach—and the threat is rising. Wheeler points out that AI‑driven patching is still years away, yet government cyber defenses are deteriorating.

The Multi‑State Information Sharing and Analysis Center lost federal funding in September 2025, and CISA has been weakened. Security researcher Nightmare Eclipse has disclosed multiple Microsoft bugs, highlighting the urgency. Public‑private partnerships like Anthropic’s Project Glasswing aim to help, but credibility suffered when access to top models was cut and deadlines missed.

Wheeler concludes that the IT director needs the shared cyber defenses once available, not an AI innovation plan.