Discord users broke into Anthropic's Mythos AI model

A group of amateur researchers on the Discord platform managed to gain unauthorized access to Anthropic's Mythos Preview artificial intelligence model, which is intended for identifying vulnerabilities in software. The company considered this tool extremely dangerous and had strictly restricted access to it. According to available information, users entered the system without complex hacking attacks, simply by analyzing open data and guessing URL addresses. Wired.com reports this.
Other serious incidents have also been recorded in the field of cybersecurity. The Citizen Lab organization reported that two spy firms, using vulnerabilities in SS7 telecommunications protocols, tracked the location of high-ranking individuals. These tracking activities, carried out via small communications operators in Israel and the United Kingdom, show that global communications systems are still vulnerable.
The US Department of Justice has announced charges against the leaders of a criminal group involved in human trafficking and cryptocurrency fraud in Southeast Asia. These groups, based in Myanmar and Cambodia, lured people with false promises of jobs and forced them to work in multi-million-dollar fraud schemes. Law enforcement agencies froze 700 million dollars related to this operation.
In the United Kingdom, it was discovered that the medical data of 500,000 citizens, belonging to the UK Biobank project, was put up for sale on the Alibaba trading platform. As a result of a breach of the terms of an agreement concluded with research institutes, genetic data and medical images fell into the wrong hands. Currently, the sale of this data has been halted, and the system access rights of the responsible organizations have been revoked.
Read “Zamin” on Telegram!