AI’s ability to see ‘mirages’ shows how alien machine brains really are

Company Developments - Anthropic has trained a new AI model named "Mythos," which is said to represent a significant advancement in capabilities, but the company is concerned about the cybersecurity risks associated with it [2] - Anthropic has experienced multiple data leaks, including sensitive information and the code for its agentic harness, raising concerns about the security of its AI models [3] - OpenAI has decided to shut down its video generation app Sora, which was unprofitable and resource-intensive, in order to focus on its core products, leading to a loss of a potential $1 billion investment from Disney [18] - Mistral AI has raised $830 million through debt financing to support the development of Nvidia-powered data centers in Europe, aiming to provide sovereign AI alternatives to U.S. tech giants [19] Industry Trends - California's governor has issued an executive order requiring AI companies to disclose safety, privacy, and bias-mitigation practices for state contracts, asserting independence from federal designations [20][21] - A study suggests that AI chatbots may help moderate political polarization by nudging users toward more centrist views, contrasting with social media's tendency to amplify extreme opinions [23][24][26] - An economist warns that countries lagging in AI adoption may face stagflation, as they could experience high interest rates without the corresponding economic growth that AI-advanced economies like the U.S. and China might enjoy [28]