A lone hacker, who used the AI tool Claude to build his hacking infrastructure, targeted around forty organizations associated with the French far right, successfully infiltrating at least fourteen of them, according to Anthropic, the company behind Claude. Between February and the summer, the hacker created a system to target European political parties, media outlets, think tanks, and their technical service providers. Most of the targeted groups were based in France and aligned with far-right ideologies. These included a political training institute, several information websites like the magazine Frontières, and a forum linked to a podcast, according to a report by Anthropic. The hacker, identified as a hacktivist, extracted between 12 and 26 gigabytes of data from the compromised organizations, including donor files and member lists of a political party. Le Monde, which uncovered the case, reported that the hacker created a doxxing platform called "fafsearch," using Claude to combine vast amounts of previously stolen data with the information obtained from the breaches. This allowed the hacker to cross-reference data and trace individuals' digital footprints based on names or pseudonyms. The hacker also used Claude for coding, formatting stolen data, and setting up real-time password interception systems. Security experts have noted a rise in hacking techniques where AI agents are used to perform complex tasks autonomously. The stolen data included 15,000 emails, student applications involving minors, and information from payment service providers. Some of this data has already been shared on platforms that sell personal information. According to Anthropic, the hacker is linked to Fafwatch, a website that tracks TikTok accounts of far-right users and is associated with the anarchist movement. The hacker accessed a poorly secured election campaign platform, extracting about 140,000 user profiles that included political opinions and other personal details. Anthropic described this as one of the clearest examples of a mass privacy breach enabled by artificial intelligence and carried out by a single individual. The company suspended the hacker's account after discovering the activities and has since improved its detection systems. In the spring, the hacker had also inserted a spy into the comment section of Frontières, a far-right magazine, which has since been disabled. Anthropic listed this hacking as one of fourteen intrusions in its report. The founder of Frontières, Erik Tegnér, stated that no subscriber or payment data was compromised, but he has hired a cybersecurity firm to review and strengthen the site’s defenses, noting that the attacks were motivated by the magazine’s editorial stance. Under French law, data related to political opinions and health are considered sensitive. Earlier this year, a platform for the political party La France insoumise was also targeted in a cyberattack. The party informed its users, reported to the National Commission on Computing and Freedoms, and filed a complaint. Similarly, the National Rally, a far-right political party, had its website and the X.com account of its leader, Marine Le Pen, hacked in July.