Cybersecurity researchers have uncovered a serious vulnerability in Base44, a popular AI-powered coding platform, that could have allowed unauthorized access to users’ private applications. The flaw has since been patched following responsible disclosure.
## The Vulnerability
The security issue was surprisingly simple to exploit. According to cloud security firm Wiz, attackers needed only a non-secret “app_id” value to access undocumented registration and email verification endpoints. This allowed them to create verified accounts for any private application on the platform.
“The vulnerability we discovered was remarkably simple to exploit,” Wiz reported. The flaw completely bypassed all authentication controls, including Single Sign-On (SSO) protections, granting full access to private applications and their data.
## How the Attack Worked
The vulnerability stemmed from a misconfiguration that exposed two critical authentication endpoints without proper restrictions:
– **Registration endpoint**: `api/apps/{app_id}/auth/register` – used for creating new user accounts
– **Verification endpoint**: `api/apps/{app_id}/auth/verify-otp` – used for email verification with one-time passwords
The “app_id” value required for these attacks was publicly visible in application URLs and manifest.json files, making it easily accessible to potential attackers. Once attackers obtained this ID, they could register accounts, verify email addresses, and gain unauthorized access to applications they didn’t own.
## Swift Response and Resolution
Following responsible disclosure on July 9, 2025, Wix (which owns Base44) implemented a fix within 24 hours. Importantly, there is no evidence that this vulnerability was ever exploited maliciously in the wild.
## Growing AI Security Concerns
This incident highlights emerging security challenges in AI-powered development tools. As these platforms gain popularity in enterprise environments, they present new attack surfaces that traditional security measures may not adequately address.
Recent security research has revealed multiple vulnerabilities in AI systems, including:
– **Prompt injection attacks** that manipulate AI models to produce malicious responses
– **Jailbreaking techniques** that bypass safety guardrails in large language models
– **Social engineering attacks** using AI tools to create convincing phishing attempts
– **Data exposure risks** through improperly configured AI systems
## The Broader Security Landscape
Security experts have identified over 1,862 Model Control Protocol (MCP) servers exposed online without proper authentication, creating additional risks for data theft and unauthorized access. These systems often store sensitive credentials like OAuth tokens and API keys, potentially granting attackers access to connected services.
## Moving Forward
As AI development accelerates, security researcher Gal Nagli emphasizes the importance of building security into these platforms from the ground up rather than as an afterthought. “The AI development landscape is evolving at unprecedented speed,” Nagli noted. “Building security into the foundation of these platforms is essential for realizing their transformative potential while protecting enterprise data.”
This incident serves as a reminder that as AI tools become more integrated into business operations, organizations must remain vigilant about emerging security threats and ensure proper security measures are in place.