… Scharre adds that AI adoption could become more complicated as systems advance from chatbots to agents capable of using software and computer networks. “Agentic AI raises this whole new set of challenges around security,” he notes. …
… Whitworth knew clever enemies would try to trick and hack the data that powered AI, the models, and the systems they sat inside. He knew AI targeting would succeed or fail based on keeping careful records about model development, and who supervised model development. …
… In Tuesday’s filing, the lawyers argued that the Pentagon “cannot simply flip a switch at a time when Anthropic currently is the only AI model cleared for use” on the department’s’s “classified systems and high-intensity combat operations are underway.” The department is working to deploy AI system… …
… As software companies and AI labs continue to train increasingly powerful AI models, they have begun announcing ambitious plans to build custom chips that serve their own specific AI needs. …
… Song notes that AI models are frequently used to grade the performance and reliability of other AI systems—and that peer-preservation behavior may already be twisting these scores. “A model may deliberately not give a peer model the correct score,” Song says. “This can have practical implications.”… …
… The world of AI is now more focused on LLMs and a different approach to training that involves feeding vast quantities of human text scraped from books, internet pages, and other sources into AI algorithms. Reinforcement learning has, however, played an important role in creating today’s AI systems. …
… Other labs, including Humans&, also aim to develop AI systems that prioritize human collaboration. Some prominent economists have called for AI researchers and companies to build systems in this way, focusing on human empowerment rather than replacement. …
… This is different from other disappearing AI products where your conversations logs often remain on other companies' servers for many months.” Meta is also adding Incognito Chat as an option in the Meta AI app. …
… It’s also another step in the company’s embrace of open-source AI models, part of a broader strategy to maintain its dominance in AI infrastructure at a time when leading AI labs are building their own custom chips. …
… Putting too much faith in AI would seem especially problematic when the tools may not behave as you expect. Agentic AI systems are particularly unpredictable because they do complex chores independently and can introduce odd errors. …