Experts found an unsecured 16TB database containing 4.3B professional records

Experts found an unsecured 16TB MongoDB database exposing about 4.3 billion professional records, largely LinkedIn-style data, which researchers say could enable large-scale AI-driven social-engineering attacks. The open database was discovered by Bob Diachenko and nexos[.]ai on 23 November 2025 and was secured two days later, on 25 November 2025, with ownership still unconfirmed.

Cybernews reported that at least three collections contained personally identifiable information, including names, emails, phone numbers, LinkedIn links, job roles, employers, work history, education, locations, skills, languages and social accounts, while nine collections were identified in total. The dataset included the profiles, unique_profiles and people collections, among others, with the unique_profiles dataset listing over 732 million records and image URLs.

Researchers noted that the leak is dangerous because such massive, structured data can be used to craft targeted phishing and CEO fraud, and to power automated social-engineering campaigns. The owner of the dataset remained unconfirmed, though a lead-generation company was suspected based on sitemap paths pointing to its website.