Google Confirms AI Helped Build Zero-Day Exploit That Bypasses Two-Factor Authentication
SecurityAdoption
Bearish

Google Confirms AI Helped Build Zero-Day Exploit That Bypasses Two-Factor Authentication

Google's Threat Intelligence Group said attackers used an AI model to discover and weaponize a zero-day flaw in a popular open-source web administration tool that bypassed two-factor authentication. Google said it worked with the vendor to patch the flaw before a planned mass exploitation campaign launched.

Nov 18, 2026, 06:01 AM3 min read

Written by CoinArticle’s AI Newsroom · from 2 cited sources. How we work

What Google Reported

Google's Threat Intelligence Group said cybercriminals used an AI model to find and weaponize a zero-day vulnerability in an unnamed, widely used open-source web administration tool. The flaw allowed attackers to bypass two-factor authentication. Google published the findings on May 11, 2026, and said the attackers were preparing a mass exploitation campaign before the company intervened with the vendor to issue a patch.

The flaw was a logic error rather than a memory bug or crash. According to Google, the software trusted a condition that skipped its 2FA check, and the AI analyzed how the software was meant to behave and found the contradiction. Google described the bypass as one that does not break the underlying encryption.

How the Exploit Was Identified

Crypto Briefing, citing the same GTIG research, described the deliverable as a Python script built to abuse the trust flaw. The script carried markers that GTIG said are rare in manually written exploits, including clean ANSI color classes, organized educational prompts, a fabricated CVSS severity score, and detailed help menus. GTIG said the code structure aligned with patterns typical of large language model output.

GTIG ruled out Google's own Gemini model as the tool used, which means the actors relied on a different AI system to both spot the vulnerability and build a working exploit. Decrypt's reporting on the same report adds that GTIG described the activity as the first time Google has confirmed AI-assisted zero-day development in the wild. Crypto Briefing's characterization of the case as the first confirmed AI-generated zero-day in the wild is consistent with that claim.

Disagreement on the Scale of the Threat

The two outlets frame the broader risk differently. Google's own report says AI lowers the barrier for adversaries and gives them expert-level force multipliers, and it links AI-generated decoy logic in malware to suspected Russia-nexus actors. Google also says China- and North Korea-linked actors are using AI to find software weaknesses. Decrypt reports that Google attributes those activities to specific state-linked groups, and Crypto Briefing summarizes the shift as a sharp drop in the skill needed to develop zero-days.

Researchers outside Google are less certain. Decrypt cites a Cambridge University study of more than 90,000 cybercrime forum threads, which found that most criminals used AI for spam and phishing rather than sophisticated exploit development, and argued that warnings about so-called Vibercriminals may overstate current disruption. Google's report and the Cambridge findings do not contradict each other on the specific case, but they point to different levels of risk across the wider criminal ecosystem.

Implications for Crypto Infrastructure

No specific cryptocurrency platform has been publicly linked to the exploit, and neither source names the affected tool. Crypto Briefing notes that 2FA is a standard control at most major exchanges, wallet providers, and DeFi front ends, and that many of these services run on or integrate open-source administration software of the type targeted here. Decrypt also notes that Google has faced its own security concerns around AI tooling, including a prompt injection flaw in its Antigravity AI coding platform patched in April.

Why It Matters

For Traders

Exchanges and wallet providers that rely on 2FA may face short-term scrutiny, so watch for security notices from platforms you use.

For Investors

Growing evidence that AI lowers exploit development costs raises the long-term security budget requirements for crypto infrastructure providers.

For Builders

Audit administrative and authentication logic for hardcoded trust exceptions, since AI tools are now effective at finding contradictions between intended and actual behavior.

This article is for information only and is not financial advice. Read the full disclaimer.

Related Articles

Latest News