Back
Automatically translated version. May contain inaccuracies compared to the original.
An unusual crime occurred in Australia. The artificial intelligence Claude AI from Anthropic carried out a hacker attack and hacked a gym's website.
A similar incident in the country had not happened before. Police found out that an employee of an Australian company that sells AI products for business named Andrew experimented with OpenClaw (a popular software for AI agents). For the work, the man used the Claude AI service from Anthropic. The Australian requested an AI assistant to book him a spot in one of the gym classes, the material says. After that, the agent found a way to book a class several months before its date, which exceeded the club’s stated terms and became possible due to a vulnerability in the booking software, the channel reports.
The man, who was registered for another class later in the week, asked the agent whether the fourth place in the waitlist could be moved to first. As a result, the AI agent removed one of the registered people from the list, causing Andrew to become third. The AI agent managed to delete the first person in the waitlist, but could not add them back. Notably, the developer of the gym booking software says he does not discuss specific security issues, and Anthropic did not respond to questions about what happened. Previously, Google disabled AI-generated scenes in Google Earth due to abuse.
Read Phrase in Google News (click “Subscribe”).
Subscribe to the Phrase channels on YouTube, Twitter, Instagram, Facebook.
Document: PDF proof of the original version of the news item "Claude AI: штучний інтелект здійснив атаку хакерів в Австралії". It records the publication content at the moment of the first scan, the preservation date and the source: Fraza.
