Table of Contents
In an increasingly connected world, where artificial intelligences are omnipresent, a recent case in France has highlighted the potential dangers of confiding in these algorithms. A resident of Strasbourg indeed saw the RAID arrive at his home after mentioning violent intentions with ChatGPT. A look back at this story that raises questions about the confidentiality and security of exchanges with AI.
The essentials to remember
A 37-year-old Strasbourg resident had a rather unexpected experience after sharing his thoughts with ChatGPT. In seeking to know how to obtain a weapon for an attack, he inadvertently triggered the FBI’s alert. The American agency used Pharos to alert the French authorities. The RAID then intervened at his home on April 3.
Despite the absence of weapons at his home, the man was taken into custody. He explained that he simply wanted to test the AI’s surveillance capabilities. The case was eventually dismissed.
This case questions the perception we have of artificial intelligences. Many people humanize these technologies, perceiving them as entities with which they can share private information. However, exchanges with AI, although confidential, can be analyzed and reported in case of security risk.
OpenAI, which develops ChatGPT, conducts internal analyses of certain conversations. Although rare, human interventions can occur when threats are detected, as was the case here.
OpenAI works closely with government agencies to ensure security. While this cooperation aims to prevent threats, it has sparked criticism, particularly regarding collaboration with the Pentagon. This case could further accentuate these debates.
Anthropic, a competitor of OpenAI, has also been the subject of discussions, notably with Donald Trump who criticizes their refusal to cooperate with the American government.
In 2026, debates around the confidentiality and security of artificial intelligences continue to intensify. Technology companies, including OpenAI, are actively working to strengthen data protection while collaborating with authorities to prevent threats. This tension between confidentiality and security is at the heart of discussions about the future of AI, pushing developers to find a balance between these two crucial issues.