NewsResearch
Microsoft: AI has cut the time from compromise to data theft from days to minutes
Microsoft's Digital Defense Report 2026, covering July 2025 to June 2026, says the median time from a vulnerability's discovery in the wild to its weaponization is now well below 24 hours, and that AI compresses post-compromise work such as credential discovery, lateral movement and data theft from days to minutes. Phishing rose to 23 percent of the intrusions Microsoft investigated, from 7 percent, and exploitation of public-facing applications to 24 percent, from 15. Microsoft cites the JadePuffer campaign of July 2026 as the first documented automated ransomware extortion attack, and says most campaigns still have a human directing them.
Why it matters
Detection and response timelines built for human attackers are now too slow. Containment has to be automatic, and patching prioritized by what is being exploited.