Archive · 3 Aug 2026
AI Cuts Vulnerability Exploitation Window to One Day, Heightens Risks
A big bank’s research team said that computers that learn from lots of data can find software problems faster than ever.
They found that the time between a problem being known and someone using it to break a system is now just one day.
Even when a fix is ready, many companies don’t apply it quickly, so about 60% of attacks happen before the fix is used.
The report says attacks are going up 18% next year, with about 75,000 attacks every hour.
But the same smart computers can also help fix problems, so they can be used for defense too.
AI models now detect thousands of previously unknown software vulnerabilities at scale.
The time between vulnerability disclosure and first exploitation has dropped to a single day.
In about 60% of breaches, a patch was already available at the time of compromise.
Cyberattacks are projected to rise 18% globally in 2025, with roughly 75,000 attacks per hour.
AI can also be used to propose code fixes and remediate vulnerabilities, offering defensive benefits.
- Who
- J.P. Morgan Asset & Wealth Management
- What
- Report on AI’s impact on cybersecurity, noting faster vulnerability discovery and exploitation
- When
- August 3, 2026
- Where
- New Delhi, India
- Why
- To warn that AI accelerates vulnerability exploitation, increasing cyber risk
- AI Vulnerability Detection Rate
- Thousands of new vulnerabilities identified in first month of testing
- Average Exploitation Window
- One day
- Patch Availability at Breach
- 60% of breaches had available patch
- Projected Cyberattack Increase 2025
- 18% globally
- Global Cybersecurity Professional Shortage
- 4.8 million