09-05-26, 03:51 PM
Hello DarkForums.
No deal. No payment. No extension.
Today I am releasing the complete data extraction from getuper.ai – an AI workflow automation platform. Over 2.26GB of raw intelligence including user prompts, exposed API keys, cloud credentials, and plaintext secrets. This leak exposes the entire backend of an AI productivity tool.
What we exploited:
- Exposed MongoDB instance with default credentials accessible via the internet
- Misconfigured S3 bucket with public read access containing backup dumps
- Environment variable file (.env) stored in the webroot with production secrets
- No encryption on user data or API traffic logging
- Insecure direct object references (IDOR) allowed us to enumerate all user records
Database Info:
- Platform: Custom Node.js + MongoDB + AWS
- Organization: GetUp AI / getuper.ai
- Total Leak Size: 2.26GB compressed .rar
- Extraction Date: May 2026
- Records Exposed: Estimated 500,000+ users and 2M+ interaction logs
Compromised Data (2.26GB):
- User database: Email addresses, hashed passwords (weak bcrypt rounds – crackable), names, subscription tiers, payment processor IDs (Stripe)
- Complete AI conversation history: Every prompt input by every user across the platform – includes personal questions, business data, API keys users submitted for integrations
- Exposed API keys: OpenAI keys, Anthropic keys, Google Cloud keys, AWS access keys – submitted by users via prompts and stored in plaintext logs
- Internal .env file: Database credentials, JWT secret, Stripe webhook secret, email server credentials
- Workflow definitions: User-created automation recipes containing internal URLs, tokens, and third-party service credentials
- Session tokens: Active JWT sessions for over 50,000 users – no invalidation observed
- Admin panel access logs: Emails and IPs of administrators with full platform access
- Stripe customer data: Partial credit card data (last 4, expiry), billing addresses, subscription histories
- Server logs: Full request/response logs including `X-Forwarded-For` headers revealing user IPs and `User-Agent` strings
- Source code backup: Entire application source (Node.js backend + React frontend) – includes business logic and unreleased features
Sample Data (representative – redacted):
[Exposed API Key from User Prompt – Log Entry]
[Internal .env File – Full Extraction]
[User Prompt Log – Sensitive Business Data]
[Workflow Definition – Internal Network Access]
[Stripe Customer Data]
Additional intelligence:
- Over 2.26GB of pure, unfiltered AI interaction data – many users unknowingly pasted their private API keys and internal company tokens, assuming they were encrypted. They were not.
- User base spans enterprises, startups, and freelancers. Some workflows include connections to internal corporate networks (CRM, internal APIs, GitHub). The SQL files include raw queries to these internal endpoints.
- Stripe secrets are live and could be used to issue refunds, cancel subscriptions, or access customer payment methods on a massive scale. The Stripe API key is scoped for all operations.
- MongoDB credentials grant full read/write access to the production database. At time of publication, the password has not been rotated.
- The AWS key pair provides access to the getuper.ai AWS account (S3, EC2, Lambda). The AWS secret is live and has not been revoked.
- The JWT secret can sign new admin tokens. Combined with the admin panel, this gives anyone complete control over the platform's user base and future updates.
Why this matters:
- 2.26GB of leaked AI prompts including sensitive user data and API keys
- Live AWS, Stripe, and MongoDB credentials – able to be used immediately for full account takeover, financial fraud, and destruction of data
- 500,000+ user accounts exposed – includes valid email addresses for targeted phishing campaigns
- The `.env` file alone allows for full compromise of the platform's infrastructure within minutes
Deliverables:
- `getuper_ai_2.26GB.rar` – the complete data set
- Folder structure:
- `/users/` – User profiles, passwords, payment info (JSON, CSV)
- `/logs/prompts/` – Complete chat histories (year‑long retention)
- `/workflows/` – Internal automation recipes and secrets
- `/env/` – Exposed environment variables and production API keys
- `/source_code/` – Full backup of the application (frontend + backend)
- `/stripe/` – Subscription data, customer records, transaction logs
Download:
Files: https://t.me/+Lv3LYdpVyJc5ZTM0
Vouch center: https://t.me/+wuOFH81f2us1YmQ0
Website: https://xorcat.deals/
No deal. No payment. No extension.
Today I am releasing the complete data extraction from getuper.ai – an AI workflow automation platform. Over 2.26GB of raw intelligence including user prompts, exposed API keys, cloud credentials, and plaintext secrets. This leak exposes the entire backend of an AI productivity tool.
What we exploited:
- Exposed MongoDB instance with default credentials accessible via the internet
- Misconfigured S3 bucket with public read access containing backup dumps
- Environment variable file (.env) stored in the webroot with production secrets
- No encryption on user data or API traffic logging
- Insecure direct object references (IDOR) allowed us to enumerate all user records
Database Info:
- Platform: Custom Node.js + MongoDB + AWS
- Organization: GetUp AI / getuper.ai
- Total Leak Size: 2.26GB compressed .rar
- Extraction Date: May 2026
- Records Exposed: Estimated 500,000+ users and 2M+ interaction logs
Compromised Data (2.26GB):
- User database: Email addresses, hashed passwords (weak bcrypt rounds – crackable), names, subscription tiers, payment processor IDs (Stripe)
- Complete AI conversation history: Every prompt input by every user across the platform – includes personal questions, business data, API keys users submitted for integrations
- Exposed API keys: OpenAI keys, Anthropic keys, Google Cloud keys, AWS access keys – submitted by users via prompts and stored in plaintext logs
- Internal .env file: Database credentials, JWT secret, Stripe webhook secret, email server credentials
- Workflow definitions: User-created automation recipes containing internal URLs, tokens, and third-party service credentials
- Session tokens: Active JWT sessions for over 50,000 users – no invalidation observed
- Admin panel access logs: Emails and IPs of administrators with full platform access
- Stripe customer data: Partial credit card data (last 4, expiry), billing addresses, subscription histories
- Server logs: Full request/response logs including `X-Forwarded-For` headers revealing user IPs and `User-Agent` strings
- Source code backup: Entire application source (Node.js backend + React frontend) – includes business logic and unreleased features
Sample Data (representative – redacted):
Quote:Quote:
[User Record]
Email: [email protected]
Plan: pro
Created: 2025-10-12
Last Active: 2026-05-01
Prompts: 1,247 interactions
[Exposed API Key from User Prompt – Log Entry]
Quote:User ID: 20439
Prompt: "Here is my OpenAI key – sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx – now help me build a sentiment analysis agent"
[Internal .env File – Full Extraction]
Quote:DB_URI="mongodb+srv://getuper_admin:[email protected]/getuper_prod"
JWT_SECRET="superSecretJWTtokenGetUp2024"
STRIPE_SECRET="sk_live_4eC39HqLyjWDarjtT1zdp7dc"
AWS_ACCESS_KEY_ID="AKIAIOSFODNN7EXAMPLE"
AWS_SECRET_ACCESS_KEY="wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"
EMAIL_HOST_PASSWORD="getupEmail2024"
ADMIN_EMAIL="[email protected]"
[User Prompt Log – Sensitive Business Data]
Quote:User ID: 57102
Timestamp: 2026-04-28T14:22:01Z
Prompt: "Analyze our Q1 sales report from our company server using this token: ghp_xxxxxxxxxxxxxxxxxxxx (GitHub token). Provide a summary."
[Workflow Definition – Internal Network Access]
Quote:User ID: 11203
Workflow Name: "Internal Report Generator"
Config: {"type":"http_request","url":"http://10.0.1.45:8080/internal/api/reports?token=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9..."}
[Stripe Customer Data]
Quote:Customer: stripe_cus_XXXXXXXXXXXXXXXXXX
Email: [email protected]
Last4: 4242
Expiry: 05/2027
Address: 123 Main St, San Francisco, CA
Subscriptions: ["pro_monthly"]
Additional intelligence:
- Over 2.26GB of pure, unfiltered AI interaction data – many users unknowingly pasted their private API keys and internal company tokens, assuming they were encrypted. They were not.
- User base spans enterprises, startups, and freelancers. Some workflows include connections to internal corporate networks (CRM, internal APIs, GitHub). The SQL files include raw queries to these internal endpoints.
- Stripe secrets are live and could be used to issue refunds, cancel subscriptions, or access customer payment methods on a massive scale. The Stripe API key is scoped for all operations.
- MongoDB credentials grant full read/write access to the production database. At time of publication, the password has not been rotated.
- The AWS key pair provides access to the getuper.ai AWS account (S3, EC2, Lambda). The AWS secret is live and has not been revoked.
- The JWT secret can sign new admin tokens. Combined with the admin panel, this gives anyone complete control over the platform's user base and future updates.
Why this matters:
- 2.26GB of leaked AI prompts including sensitive user data and API keys
- Live AWS, Stripe, and MongoDB credentials – able to be used immediately for full account takeover, financial fraud, and destruction of data
- 500,000+ user accounts exposed – includes valid email addresses for targeted phishing campaigns
- The `.env` file alone allows for full compromise of the platform's infrastructure within minutes
Deliverables:
- `getuper_ai_2.26GB.rar` – the complete data set
- Folder structure:
- `/users/` – User profiles, passwords, payment info (JSON, CSV)
- `/logs/prompts/` – Complete chat histories (year‑long retention)
- `/workflows/` – Internal automation recipes and secrets
- `/env/` – Exposed environment variables and production API keys
- `/source_code/` – Full backup of the application (frontend + backend)
- `/stripe/` – Subscription data, customer records, transaction logs
Download:
Files: https://t.me/+Lv3LYdpVyJc5ZTM0
Vouch center: https://t.me/+wuOFH81f2us1YmQ0
Website: https://xorcat.deals/
