AI is increasingly writing the world’s code—but with speed comes risk. A new study finds that nearly one in five serious security breaches now originate from AI-generated code, and almost 70% of companies have discovered vulnerabilities linked to machine-written software.

Published in Aikido Security’s State of AI in Security & Development 2026, the research surveyed 450 developers, security leaders, and application engineers across the U.S. and Europe, highlighting how AI-driven programming is advancing faster than the safeguards meant to protect it.

Figure 1. AI Speed Boost May Leave Systems Vulnerable to Hackers.

Speed Brings New Risks

The study found that 24% of production code worldwide is now AI-generated—29% in the U.S. and 21% in Europe. While this has boosted productivity, it has also introduced new security challenges.

• 1 in 5 companies reported a serious breach caused by AI code.
• Another 49% experienced smaller issues or only realized problems later.
• Ambiguity over responsibility is common: 53% blame security teams, 45% blame the developer using AI, and 42% blame the person who merged the code.

This uncertainty slows fixes, prolongs investigations, and leaves vulnerabilities open longer than ideal.

Regional Differences

European firms tend to be more cautious, reporting only 20% of major breaches linked to AI code, compared with 43% in the U.S. The difference reflects stricter European regulations and a more measured approach to automation, whereas American developers often prioritize speed over safeguards. Figure 1 shows AI Speed Boost May Leave Systems Vulnerable to Hackers.

Interestingly, U.S. teams are more proactive in tracking AI-generated content: nearly 60% log and review every line, compared to just over a third in Europe. This gives U.S. firms greater visibility but also exposes them to more risk.

Tool Overload and Productivity Costs

Many companies juggle multiple security tools, but more tools don’t always mean better security. Teams using 1–2 security products fixed critical issues in about 3 days, while those using 5 or more took nearly 8 days.

False positives further hinder productivity. Engineers spend hours each week sifting through harmless alerts, sometimes disabling scanners just to ship code—often introducing hidden risks [1]. One respondent summarized it as “too many alarms and not enough clarity,” a sentiment echoed across regions.

Humans Still Crucial

Despite AI’s growing role, human oversight remains vital:

• 96% of respondents believe AI will eventually write secure code, but most expect it will take at least five years.
• Only 20% think AI will do so without human review.
• Experienced security engineers remain key; 25% of CISOs warn that losing a single skilled team member could directly lead to a breach.

Optimism remains high, though: 90% of firms expect AI will soon handle most penetration testing, and nearly 80% already use AI to help fix vulnerabilities. The challenge lies in balancing automation with human judgment.

conclusion: Slow Down to Stay Safe

Aikido’s report underscores a simple truth: the faster AI generates code, the faster mistakes can spread. Security still depends on developers who understand what the AI is doing and take responsibility for the results.

In short, while AI is racing ahead, the guardrails are still catching up. For now, slowing down to double-check machine-written code may be the smartest approach.

reference:

1. https://www.digitalinformationworld.com/2025/10/ais-shortcut-to-speed-is-leaving-door.html

Cite this article:

Keerthana S (2025), AI’s Fast-Track Approach to Efficiency Could Expose Systems to Hackers, AnaTechMaz, pp.844