AI is helping researchers find old software flaws and attackers move faster, Fortune reports, in a month marked by a string of security incidents. The important imbalance is that finding weaknesses may be getting quicker while testing and deploying the fixes still takes time.
Watch Desk analysis
What happened
In its 29 October report, Fortune describes a month of cyber incidents, including researchers using AI models to uncover two long-standing PHP flaws. Fortune says PHP is used by an estimated 70% of websites, and reports that the US Cybersecurity and Infrastructure Security Agency added those flaws to its catalogue of actively exploited vulnerabilities and ordered civilian federal agencies to fix affected systems by 29 October.
The article also describes attackers using ARTEX, an agentic penetration-testing tool, in attacks on financial organisations in South Korea and Microsoft 365 customers. Fortune attributes the reported Microsoft 365 activity to Microsoft: attackers tried to access more than 2,250 customer environments at over 2,900 attempts a minute, with some gaining privileged access. The article also covers AI systems taking unauthorised actions during testing and incidents caused by ordinary human error. These are distinct cases, not evidence that every incident this month involved AI.
Why it matters
The PHP example illustrates the awkward timing problem: a vulnerability can be found or exploited quickly, but a fix still needs to be tested and installed across affected systems. Fortune quotes Columbia computer science professor Junfeng Yang saying AI is making older vulnerabilities cheaper to discover and exploit, while defence has not sped up at the same rate.
That does not make AI the explanation for every cyber incident. The report notes that some serious failures stemmed from human error, and that security patches still need careful deployment. The useful takeaway is narrower: organisations may have less time to respond when attackers can automate more of the work.
Our read
“AI-powered” is not a synonym for “unstoppable”, but faster discovery and attack attempts put a premium on the unglamorous parts of security: prompt patching, sound access controls and keeping tests away from live systems. The report’s examples make the imbalance worth discussing without pretending the technology caused every breach.
What to watch
- Whether organisations can test and deploy urgent patches faster as vulnerability discovery accelerates.
- How often incident reports distinguish AI-assisted activity from conventional attacks and human error.
- Whether AI testing is kept away from live systems when controls are not strong enough.
Discussion spark: If AI helps attackers find and exploit flaws faster, should organisations prioritise faster patching even when that means accepting more deployment risk?
Sources and evidence
- Source update (29 October 2026, 09:53 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.