AI offensive cyber capabilities are doubling every six months, safety researchers find

THE DECODER / 4/5/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

Key Points

  • A new safety research study finds that AI-driven offensive cyber capabilities have been improving rapidly, doubling roughly every 5.7 months since 2024.
  • The researchers report that models such as Opus 4.6 and GPT-5.3 Codex can now solve certain exploitation tasks that typically take human experts about three hours.
  • The study highlights accelerating progress in AI-assisted vulnerability exploitation, raising concerns for cyber defense and safety planning.
  • These findings suggest organizations may need to update threat models, detection strategies, and red-teaming practices to account for faster attacker capability growth.

AI models are rapidly improving at exploiting security vulnerabilities. According to a new study, their offensive cyber capability has been doubling every 5.7 months since 2024, with Opus 4.6 and GPT-5.3 Codex now solving tasks that take human experts about three hours.

The article AI offensive cyber capabilities are doubling every six months, safety researchers find appeared first on The Decoder.