Accra:Meta has suspended the conversation-sharing feature in its Muse artificial intelligence (AI) assistant following revelations about potential privacy breaches. Cybersecurity researchers discovered that thousands of shared chat links were accessible via public internet records, posing a risk of exposing sensitive personal and corporate data.

According to Wam, cybersecurity firm Claroc identified over 6,800 shared links through passive DNS records. These records, which log domain-name lookups, can render web addresses discoverable, thereby compromising user privacy.

Researchers were able to access several hundred conversations that included sensitive information such as corporate revenue reviews, loan applications, bank fraud alerts with partial account details, and confidential job interview notes. This vulnerability was linked to Muse's URL design, which incorporated conversation titles and randomly generated identifiers. Once accessed, these links could be logged in public databases, enabling others to find them without needing to receive them directly. The conversations could be viewed without signing in.

The incident underscores the escalating concerns surrounding data privacy in AI applications and the inherent risks of sharing conversations that may contain confidential information.