Home Page
Results if no vulnerabilities are found
Results if vulnerabilities are found

🛡️ AICert: AI Security Certification Platform

Submission of Group 24 for HackEurope 2026, Security Track

Why This Matters

AI-powered applications are being shipped faster than ever. Most developers are experts in building features, not in adversarial machine learning or prompt security. The result: a growing ecosystem of LLM-powered products with no standardized security baseline, no audit trail, and no way for users or customers to know if an app has been hardened against AI-specific attacks.

Meanwhile, the threat landscape is accelerating. A college student with API access can now probe an LLM app for prompt injection, jailbreaks, and data exfiltration in minutes. Yet our defensive infrastructure remains fragmented and reactive.

This is the core insight behind BlueDot Impact's Defensive Acceleration philosophy: the best response to AI-enabled threats is not slowing down AI development, it is accelerating the defensive tools that keep pace with it.

AICert is our contribution to that mission. By giving developers a fast, automated, and standardized way to audit and certify their LLM applications, we create an incentive structure for security to be built in from the start, not bolted on after an incident. A certification badge in a README or product page is a small thing. The behavior change it drives is not.

Overview

AICert is a platform that analyzes codebases for AI/LLM security vulnerabilities and issues verifiable security certifications based on the OWASP Top 10 for LLM Applications.

Companies can submit their AI-powered projects for automated security scanning. The platform evaluates code against industry-standard security categories, including prompt injection, insecure output handling, sensitive information disclosure, and more, and assigns a certification tier (🥉 Bronze, 🥈 Silver, 🥇 Gold) based on the results.

Key Features

Automated AI Security Analysis — Submit a GitHub repository and receive a detailed security audit
OWASP LLM Top 10 Coverage — Scans for all major AI/LLM vulnerability categories
Certification and Badges — Earn verifiable certifications with embeddable badges
Public Verification — Anyone can verify a certification via a unique verification ID
Real-time Scanning — Live progress tracking during analysis

Tech Stack

Frontend: React, TypeScript, Tailwind CSS, Framer Motion
Backend: Lovable Cloud (authentication, database, edge functions)
AI Engine: Google Gemini (under the hood) for deep code security analysis, including static pattern detection, semantic reasoning over code structure, and natural language generation of findings and remediation advice
Built with: Lovable

How It Works

Sign up and create a company profile
Submit a project with a GitHub repository URL
Under the hood, the platform sends the repository to Google Gemini, which performs deep semantic analysis of the code against each OWASP LLM Top 10 category
Receive a detailed report with findings, severity ratings, and remediation advice
If the score meets the threshold, earn a verifiable security certification

Certification Tiers

Tier	Score	Badge
Gold	95+	🥇
Silver	85+	🥈
Bronze	70+	🥉

Certificates are publicly verifiable at a unique URL so anyone can confirm authenticity.

OWASP LLM Top 10 Coverage

AICert evaluates submissions against all ten categories from the OWASP Top 10 for LLM Applications:

#	Category
LLM01	Prompt Injection
LLM02	Insecure Output Handling
LLM03	Training Data Poisoning
LLM04	Model Denial of Service
LLM05	Supply Chain Vulnerabilities
LLM06	Sensitive Information Disclosure
LLM07	Insecure Plugin Design
LLM08	Excessive Agency
LLM09	Overreliance
LLM10	Model Theft