Anthropic's Claude Mythos: Powerful AI for Defense, Not Public

0:00It's April 9th. I'm Corey with The Story. This is Calgary News Today, driven by AI, andthropic drop word on their latest AI beast, Claude Mythos. Part of the Claude family like those chatbots you know. This frontier model crushes software engineering, reasoning, computer tasks, knowledge work, and research leaps ahead of anything they've built before. But its killer cyber security chops, for fixing buds or exporting them, make it too risky for public hands. So they're holding it back from everyone, no general release. Instead, it's going to a tight crew of partners who run critical software setups, lock down for peer defense against cyber threats. Experts are buzzing with real talk. A senior researcher flags legit national security worries that ain't just talk. We gotta watch close. A CEO calls it a smart conscious move to shield infrastructure, but best someone else will cook up the same power and sell it wide. Meanwhile, Canada's cyber scene is heating up bad.

1:01Ransomware hit an average 26% yearly rise from 2021 to 2024, with recovery costs doubling to 1.2. Billion dollars in 2023, AI is making these attacks cheaper, faster and sneakier for businesses, big and small. As governments lag on AI rules, firms like Anthropic are drawing lines on what's safe to drop. This models a wake up on balancing breakthroughs with keeping our digital world locked tight.