Cybersecurity researchers from Calif used the widely discussed AI model Claude Mythos to bypass Apple’s Memory Integrity Enforcement, or MIE, protection mechanism in macOS.
By Chris Borden OpenAI is also reportedly working on a new AI model with advanced cybersecurity capabilities that would be made available only to a small group of companies, according to Axios. The approach mirrors Anthropic’s, which on Tuesday restricted access to its new model, Mythos Preview, to selected technology and security firms because of its unusually strong hacking capabilities.
By Daniel Mercer Anthropic has developed a new model called Claude Mythos, but chose not to release it publicly because of serious security concerns.
By Daniel Mercer “AI can make mistakes, please verify responses” appears under every AI-generated search answer from Google. However, how often these errors actually occur has rarely been systematically studied.
By Alex Rowland OpenAI has proposed a plan to reshape public policy in the age of “superintelligence.” Artificial intelligence will transform banking, the labor market, and entire sectors of the global economy. JPMorgan Chase CEO Jamie Dimon said this in his annual letter to shareholders.
By Alex Rowland Politico reports that the European Commission, Parliament, and Council have banned their press teams from using fully AI-generated content. Experts say the decision represents a missed opportunity.
By Alex Rowland AI agents inherit the weaknesses of large language models, but their autonomy and access to external tools create a fundamentally new attack surface. A Google DeepMind paper introduces this attack surface.
By Chris Borden The White House has released a national AI framework along with legislative recommendations for the US Congress. The document outlines seven key focus areas.
By Daniel Mercer Media analysis firm Newsguard, together with AI detection company Pangram Labs, has launched a real-time system designed to identify so-called AI content farms. More than 3,000 websites have already been flagged, and the number is rising quickly.
By Alex Rowland OpenClaw’s rapid surge in popularity has prompted a warning from China’s National Computer Network Emergency Response Technical Team/Coordination Center (CNCERT/CC), which flagged multiple security risks associated with the AI-agent platform.
ES 
EN