A significant incident in Moscow has highlighted the dangers of using AI networks for sensitive work information. A top executive at a major engineering firm was dismissed for leaking confidential documents, including uploading them to a Chinese AI network known as DeepSeek. The executive, who held the position of sales director for less than six months and earned a salary exceeding 800,000 rubles per month, was found to have repeatedly violated company confidentiality protocols.
The situation unfolded when the employee forwarded internal emails to her personal email address, cleverly using the blind carbon copy feature to keep it hidden from her superiors. Additionally, she uploaded proprietary internal materials to DeepSeek, ostensibly seeking help from the AI for her tasks. However, this act resulted in the company's sensitive data being stored on an unsecure external server.
The court ruled in favor of the employer, stating that the executive had no legitimate work-related reason to transfer information from secure local sources to an external address. The court deemed her actions a serious breach of her employment duties. Following the leak, one of the company's partners ceased communication during negotiations, contributing to unmet sales targets.
In her quest for a golden parachute of 5 million rubles and a change in her termination's wording to "mutual agreement," she turned down an amicable settlement of over 400,000 rubles. Ultimately, the court rejected all her claims and upheld the legality of her dismissal.
This case is not an isolated incident; it raises concerns about the potential for data leaks through AI networks. Instances of sensitive information inadvertently becoming publicly accessible have been reported before. For example, in July 2026, user conversations from DeepSeek were indexed by Google due to improper access settings. Similarly, OpenAI's ChatGPT had a feature that allowed users to make their chats visible in search engines, leading to the indexing of thousands of personal and sensitive discussions.
The common thread in these cases is the failure of developers to secure shared content from indexing by search engines. Users must be cautious about what they upload to AI networks, recognizing that anything shared could potentially be accessed by others.
The key takeaway from this incident is clear: using AI networks does not equate to secure data handling. Users must treat interactions with these technologies as public conversations, ensuring that they do not disclose confidential or sensitive information. As the market for AI continues to grow, companies and individuals must prioritize data privacy and security to avoid similar pitfalls. This situation serves as a cautionary tale for competitors, emphasizing the need for robust protocols and user education regarding the risks associated with AI technologies.
Informational material. 18+.
