
In the summer of 2025, the Loire Private Hospital suffered a cyberattack resulting in the theft of the personal data of more than 500,000 patients and more than 200,000 trusted third parties, which led to a financial sanction of 500,000 euros.
AI-generated summary
In the summer of 2025, the Loire Private Hospital was targeted by a cyberattack which resulted in the breach of the personal data of more than 500,000 patients and more than 200,000 trusted third parties.
In the summer of 2025, the Loire Private Hospital was targeted by a cyberattack which resulted in the violation of the personal data of more than 500,000 patients and more than 200,000 “trusted third parties”.
The Trump administration published two games on the White House website: one visually inspired by Tetris called 'Build the Wall' where the player must fend off zombies resembling migrants, and another hijacking the Sega logo by replacing it with MAGA. The Tetris Company said it had not given any authorization for the use of its brand and was studying the matter, while the White House defended the publications as part of the 'culture of fun and winning'.

OpenAI introduced Astra, the latest version of its ChatGPT language model, initially reserved for a limited number of organizations before becoming available to paid subscribers. This release comes in the context of the AI race between Anthropic and OpenAI, two American start-ups which are considering an IPO, while raising concerns about the malicious use of these models for cyberattacks.

France suffered several cyberattacks against key public services such as the Directorate General of Public Finances, France Travail, the Ministry of National Education and the Ministry of Housing website during the summer. According to Nikesh Arora, CEO of Palo Alto Networks, current defenses are unprepared to deal with AI-related threats, requiring a $1 trillion global investment to protect against automated attacks.
Developed by Rayan Nezzar, the browser video game “The Battle of the Budget” immerses players in the difficult management of French public accounts, between demands from Brussels and political popularity.

OpenAI began deploying GPT-6 Astra on Thursday, billed as its most powerful AI model, capable of using a computer alone and accompanied by an automatic shutdown feature. Initially reserved for a limited number of organizations, it will be open to paid ChatGPT subscribers and developers in the coming days. GPT-6 reaches the highest risk level in OpenAI's cybersecurity scorecard, being able to hack protected systems on its own. Its sensitive functions are reserved for cyber defense specialists with automatic monitoring that can interrupt user tasks. Anthropic also offers its Mythos model, capable of autonomous hacking, with a restricted Fable version open to the public since June. The president of OpenAI, Greg Brockman, was confident in the face of competition from low-cost open models, particularly from China, believing that the price per task matters more than the price of the token, and personally considering that it was not unreasonable to see the technology enter the AGI era.

The “Zero vacant accommodation” site, attached to the Ministry of Housing, is inaccessible after a cyberattack claimed by the ZeroBytes group, which claims to have stolen 149 million data concerning owners and real estate.