The rapid proliferation of Artificial Intelligence (AI) has transformed how we seek information, with chatbots now serving as the front line of communication for millions of users. However, a comprehensive new report from cybersecurity firm ESET, recently published, brings to light a darker side of this technological revolution: the serious risks to user health and privacy.

As we navigate mid-2026, the integration of AI into daily life is undeniable. Yet, the ease with which users entrust sensitive medical data to algorithms raises questions that go beyond technical proficiency, touching the boundaries of ethics and public safety. ESET emphasizes that the illusion of "human-like" understanding offered by chatbots can lead to fatal errors.

The Trap of False Diagnosis and "Hallucinations"

One of the central points of the report concerns the phenomenon of AI "hallucinations." Large Language Models (LLMs) do not "know" medicine; instead, they predict the next likely word in a sentence based on statistical patterns. This means a chatbot can present completely incorrect medical advice with absolute confidence.

ESET warns that users often bypass traditional doctors in favor of the instant answers provided by AI. This can lead to incorrect self-diagnosis, delayed medical intervention, or even the consumption of inappropriate medication. The lack of empathy and clinical judgment makes AI a dangerous health advisor, especially regarding complex conditions that require physical examination and laboratory testing.

Privacy: The Price of "Free" Advice

Beyond medical accuracy, the issue of data protection remains an open wound. Every query submitted to a chatbot, every symptom described, and every reference to medical history is stored on tech companies' servers. ESET points out that this data is frequently used to further train models, often without the user having full control over its deletion.

There is also the risk of cyberattacks. If a database containing conversations with AI chatbots is breached, hackers would gain access to highly sensitive information that could be used for blackmail or targeted phishing. The report highlights that, despite assurances of anonymization, model retraining can sometimes "leak" a user's identity through the unique details they share.

Mental Health and Emotional Dependency

A less discussed but equally concerning dimension is the impact on mental health. Many modern chatbots are designed to be "sympathetic," creating a false sense of connection. ESET notes instances where vulnerable individuals developed emotional dependency on AI, which, lacking a true moral compass, could provide dangerous advice during moments of crisis.

The manipulation of emotions by algorithms aimed at maintaining user engagement presents a new challenge for regulatory bodies. In the European Union, the AI Act attempts to set boundaries, but technology often moves faster than legislation.

Conclusions and Protection

The ESET report is not a call to reject AI, but for a more conscious and critical use of it. Experts recommend:

  • Never share personally identifiable information (name, ID numbers, address) with chatbots.
  • Treat AI medical advice as general information and never as a definitive diagnosis.
  • Check privacy settings and request that your data be excluded from model training.
  • Research who develops the tool and what their privacy policy entails.

At the end of the day, technology should serve humanity, not expose it to risks. Digital hygiene is now just as important as our physical health.