Using artificial intelligence to help write code might sound like a smart shortcut, but new research shows it’s creating big security headaches. A security firm called Apiiro studied thousands of developers and their code repositories and found that those using AI produced ten times more security problems than those who didn’t. The faster pace of coding with AI seems to be the main issue, as developers are generating more code at a rapid rate, which often leads to mistakes or vulnerabilities.

AI’s Double-Edged Sword in Coding

AI can make coding faster and reduce common errors like syntax mistakes and logic bugs. Apiiro’s research found that syntax errors dropped by 76 percent and logic bugs decreased by 60 percent when AI was involved. Sounds good, right? But there’s a major catch. While some errors are being caught, new and more dangerous problems are popping up. Privilege escalation, which allows hackers to gain higher access to a system than they should, increased by an astonishing 322 percent. Architectural design issues also rose sharply, up 153 percent.

The Hidden Costs of AI in Security

This pattern suggests that AI might be fixing small problems but creating bigger risks. Apiiro’s product manager, Itay Nussbaum, explained that AI is like “fixing the typos but creating the timebombs.” The security gaps aren’t just minor—they could be severe enough to give attackers a foothold inside systems. Researchers from the University of San Francisco, Canada’s Vector Institute, and the University of Massachusetts Boston also found that AI’s improvements tend to weaken overall security. That’s a big concern, especially as companies like Coinbase, Shopify, and Duolingo now require their employees to use AI tools to code.

More Problems for Developers and Businesses

As AI becomes more integrated into workplaces, the amount of vulnerabilities and the workload for security teams grow. Many companies are finding themselves spending more time fixing issues created by AI-assisted coding. Instead of reducing work, AI appears to be shifting the burden onto human workers, who now have to deal with more security flaws and architectural problems. The trend isn’t slowing down, and experts warn that security risks could get even worse as AI continues to be adopted more widely in coding and other creative fields.

In the end, it’s clear that AI in coding is a mixed bag. While it speeds up development and reduces simple errors, it also introduces new, serious vulnerabilities that demand extra attention. For businesses and developers, understanding these risks is key to making smarter choices about how and when to use AI tools. The challenge will be balancing the efficiency gains with the need for security and safety in our increasingly digital world.

Inspired by

• https://futurism.com/ai-coding-security-problems

Sources

• gmpg.org
• apiiro.com
• theregister.com
• arxiv.org
• forbes.com