Mythos AI: Australia Seeks Access to 'Dangerous' Model

Australian agencies and businesses are seeking access to a new AI model called Mythos, developed by AI firm Anthropic. The chief lawyer for Anthropic is in Australia for meetings with senior government officials, discussing this access. Mythos is considered "dangerous" because of its capability in finding cybersecurity weaknesses. It has only been released to about 50 mostly US-based companies, with concerns it could be used for cyber-attacks if it falls into the wrong hands. Organisations like the UK's AI Safety Institute and Mozilla have tested and confirmed Mythos's advanced capabilities. The Australian government is keen to attract top AI companies to establish a presence in the country, even considering reviewing copyright laws as part of a deal. This push highlights the growing importance of advanced AI models for national security and industry.