"Dark Sourcery" AI Chatbot Answer Poisoning Scam

Overview

Scammers are flooding the web with fake posts, PDFs, reviews, and support pages specifically engineered to get AI chatbots like ChatGPT, Google Gemini, and Google AI Overview to repeat fraudulent customer support phone numbers and login pages as if they were official company contact details.

How It Works

Researchers at Vigilance Security identified a campaign they call "Dark Sourcery" in which attackers borrow search engine optimization techniques, but apply them to AI answer engines instead. They mass-produce content such as fake forum posts, PDFs, reviews, and support pages seeded with fraudulent phone numbers, email addresses, and login pages, designed to be picked up and repeated by AI assistants when users ask "what is the customer support number for [company]." When a victim asks a chatbot for a company's support line, refund department, account recovery help, or software download link, the AI may confidently present the attacker's fabricated contact information as fact, without the user ever visiting the poisoned page that supplied it. Victims who call the fake number or visit the fake login page are then asked for credit card numbers, bank login credentials, or other payment details by someone posing as legitimate company support staff.

Red Flags

• A phone number or contact link for a well-known company was provided by an AI chatbot or AI search summary rather than the company's own verified website
• The 'support agent' who answers immediately asks for a credit card number, full bank login, or payment card data to 'fix' an account, refund, or booking
• The support contact info surfaced by the AI cannot be found on the official company website or app when checked directly
• Unsolicited offers to help move a flight, unlock a bank account, or process a refund come from a number obtained via a chatbot search rather than a number you looked up yourself

Real Examples

• Vigilance Security researchers counted tens of thousands of malicious pages seeded with fake support numbers, email addresses, login pages, and software-update instructions, affecting at least 374 companies including Fortune 100 firms, airlines, banks, travel companies, and software providers.
• Named brands caught up in the campaign reportedly included Delta, Lufthansa, Qatar Airways, Chase, Bank of America, Airbnb, and TripAdvisor.
• Vigilance told Dark Reading it had seen a few occurrences of people complaining about being scammed into giving up payment card details after a chatbot-provided phone number connected them to someone who offered to help move a flight or unlock a bank account before asking for credit card details.

Where It Spreads

How to Protect Yourself

• Never call a support number or click a login link that an AI chatbot or AI search summary gives you — go directly to the company's official app or website and find contact details there instead
• Independently verify any phone number, email address, or URL for customer support, refunds, account recovery, or software downloads against the company's verified official channels before using it
• Be especially cautious with AI-provided contact details for sensitive tasks like account recovery, payments, refunds, or software updates, since these are the exact categories attackers are targeting
• If you already called a number from a chatbot answer and gave out payment or account information, contact your bank or card issuer immediately to dispute charges and freeze the account

Related Scams