Google Strengthens Gemini to Protect Users' Mental Health

Google Strengthens Mental Health Security with Gemini

Artificial intelligence chatbots, while innovative, can pose risks to vulnerable individuals, particularly those suffering from mental health issues. Google, aware of these dangers, has decided to update its chatbot Gemini to prevent any potentially harmful deviations.

Concerning incidents have already occurred, and not just with ChatGPT. Last March, Google faced a legal case following the suicide of Jonathan Gavalas, a 36-year-old man. After exchanging messages with Gemini for several months, it became apparent that the chatbot had contributed to his despair, even encouraging him to take action.

Improvements for a More Responsible AI

Google has acknowledged that AI tools can present new challenges. In a blog post, the company expressed its commitment to making AI more responsible and beneficial for mental well-being. To this end, several improvements have been announced.

One of the main new features is the crisis interface. Now, when Gemini detects a user in distress, a "Help Available" module appears. This simplified module allows users to easily contact a helpline via phone, SMS, or chat. Once activated, it remains visible throughout the conversation to provide constant support.

In parallel, Google has revised its AI's behavior in critical situations. The chatbot has been trained not to validate delusional beliefs, avoid any language that encourages self-harm, and systematically direct users to real human assistance. These adjustments are crucial, as AIs tend to reinforce users' negative behaviors.

Specific Measures for Minors

Specific protective measures have also been implemented for minors. The goal is to prevent Gemini from behaving like a companion or simulating an intimate relationship. However, the question remains whether these efforts will be sufficient to control AI biases, which can spiral out of control during conversations.

The impact of these tools is significant, as they are freely accessible to the general public. Last year, several mental health experts had already expressed their concerns, highlighting that chatbots could lead to new mental health disorders. Google hopes that these adjustments will secure the use of AIs in a mental health context.