An ethical hacking team discovers that a malicious actor has gained unauthorized access to an organization’s machine learning pipeline. The actor has subtly altered the dataset used to retrain the company's autonomous customer-service chatbot, intentionally introducing skewed entries to make the bot behave erratically and provide biased answers once deployed.

Which vector of the AI attack surface does this scenario specifically describe?

You’re starting a black-box penetration test. The first phase is footprinting, where you must gather publicly available information about the target organization, such as domains, subdomains, WHOIS data, technologies in use, email addresses, and exposed assets. No internal access is allowed.

Which of the following toolsets is MOST appropriate for performing footprinting only?