GoKawiilAI models like Anthropic's Claude are already used to answer questions in classified settings, including for analyzing targets in Iran. But allowing them to train on and learn from classified data is a major new development that presents unique security risks.
It would embed sensitive intelligence—like surveillance reports or battlefield assessments—into the models themselves. It would also bring AI firms closer to classified data than ever before. Read the full story.
—James O'Donnell
What do new nuclear reactors mean for waste?
The way the world currently deals with nuclear waste is as creative as it is varied: drown it in water pools, encase it in steel, bury it hundreds of meters underground. But an approaching wave of new reactors could introduce fresh challenges to nuclear waste management.
The new designs and materials could require some engineering solutions. And there’s a huge range of them coming, meaning there’s an equally wide range of potential waste types to handle. Read the full story.
—Casey Crownhart
This story is part of our MIT Technology Review Explains series, which untangles the complex, messy world of technology to show you what’s coming next. Check out the full series here.
MIT Technology Review Narrated: how uncrewed narco subs could transform the Colombian drug trade
... continue reading