371
Productivity & Workflow357
Software Development253
Automation & Workflow230
Marketing & Growth213
AI Infrastructure & MLOps197
Photography & Imaging170
Writing & Content Creation197
Design & Creative176
Data & Analytics162
Customer Support161
Education & Learning150
Sales & Outreach132
Voice & Speech131
Operations & Admin118
Researchers are using AI to find long-standing bugs faster than teams can fix them, raising the risk that attackers exploit systems before patches land.
In short: AI is helping researchers find hidden security problems in widely used software much faster, but fixing those problems still takes time.
AI tools are making it quicker to scan big codebases for weaknesses. Code is the instructions that make software run, and a weakness can act like an unlocked door that lets an attacker get in.
Recent research shows how large the gap can be between finding and fixing issues. Microsoft researchers used AI-assisted scanning on parts of the Linux kernel, which is the core of the Linux operating system (like the engine in a car). In about two weeks, the scan produced 26,502 possible issues, and human reviewers narrowed that list to 59 reported vulnerabilities, including eight with working “root” exploits, meaning they could give an attacker the highest level of control.
Researchers also used AI tools to uncover serious flaws in PHP, software that helps run many websites. Some of these flaws could enable “remote code execution,” which means an attacker can run their own commands on a server from far away (like controlling a computer without touching it). Reports say attackers started trying to use the PHP flaws after details leaked before fixes were widely available.
A similar pattern showed up in Java, where an AI-assisted find in a widely used component suggested that a single bug in a common building block can put many apps at risk.
The slow part now is cleanup, not discovery. Teams must confirm a flaw, coordinate a fix, test it, and roll it out without breaking systems. The risk is a race, defenders and attackers can both use AI to move faster, so organizations may need better inventories of what software they run, and quicker patch plans for internet-facing and critical systems.
Source: NYTimes