AI-powered hacking has taken a worrying shift. Google has found a system that can find software flaws and launch attacks on its own. The discovery raises fresh questions about online safety, fast security updates and the future of cybercrime.
A hacker detected by Google that works on its own
As reported by The Times of India, Google says it has found the first known case of hackers using an AI system to find and exploit software flaws without a person guiding each step. The tool is called exploit_agent. It was found on an exposed test server and links to several large language models, or LLMs. The system checks changes in software code before and after a security patch. A patch is meant to fix a flaw. But by comparing the old and new code, the AI can work out what weakness was fixed.
The server also held data taken from victims. Google found scripts that could help attackers search inside a hacked network, guess passwords, install webshells and remove tools left by rival hackers. A webshell is a small tool that lets an attacker control a hacked server. The operator appeared to show interest in French websites linked to media, entertainment and education. However, Google’s team did not confirm that exploit_agent itself carried out the data theft.
Google: Why security updates may not be enough
The discovery matters because attackers may now have less time to act after a flaw becomes public. Security teams often release patches so users and firms can fix weak software. Yet attackers can study the same patch to learn what was wrong. These are known as n-day flaws. They are not secret bugs. A fix is already out, but many users may not have installed it. AI tools can help hackers turn such flaws into working attacks much faster.
Google has seen attackers use AI frameworks such as ARTEX, CAIRN, CyberStrikeAI and PentAGI. One group used a PentAGI-based system to scan firms across Asia. CAIRN was seen in an attack on the US aerospace sector. Google also pointed to a campaign linked to flaws in PaperCut print servers. Researchers believe AI helped build and test exploits.

What this means for users and businesses
The findings do not mean that every phone or laptop is now at risk from an AI hacker. They do show that the time between a flaw being found and an attack being launched may be getting shorter. Google says these systems create activity that may help security teams spot them. But their speed and scale still pose a challenge. Hackers use known methods, while AI lets them carry out more steps at once.
For ordinary users, the advice remains simple. Install phone, laptop and app updates when they are available. Use strong, unique passwords and turn on multi-factor authentication where possible. Businesses should also track devices, limit access to key systems and fix serious flaws quickly. AI-led attacks have grown in just one quarter. Old software gaps can become dangerous when machines help attackers move faster.
Author’s Opinion
AI-powered hacking is a serious concern, but panic is not the answer. The bigger challenge is speed. Companies must fix flaws faster, while users must keep their devices updated. Good security habits should remain the first line of defence.
READ MORE: Meta fined €805m over Facebook accessing users’ photos.
ALSO READ: SpaceX makes history with Starship after booster loss.
READ NEXT: Microsoft restores Outlook and Teams after global outage.


Comments
0Be the first to join the discussion.