"We already know that Musk’s people have access to large swaths of information from federal agencies—what we don’t know is what they’ve copied, exfiltrated, or otherwise taken with them. In theory, this material, whether usable together or not, could be recombined with other identifying information from private companies for all kinds of purposes. There has been speculation already that it could be fed into third-party large language models to train them or make the information more usable (Musk’s xAI has its own model, Grok); outside firms could use their own technologies to make sense of disparate sets of data, as well. Such approaches, the federal workers told us, could make it easier to turn previously obfuscated information, such as the individual elements of a tax return, into something to be mined."

"The thought that the government would centralize or even give away citizen data for private use is scandalous. But it’s also, in a way, expected. The Vietnam War and Watergate gave Americans reasons to believe that the government can’t be trusted. The Cold War issued a constant, decades-long threat of annihilation and the necessary surveillance to avoid it. The War on Terror extended the logic into the 21st century. Optical, recording, and then computer technologies arose, offering new ways to watch the public. During the 2010s, Edward Snowden’s NSA surveillance leaks took place, and the Facebook–Cambridge Analytica scandal was brewing. By then, the 20th-century assumption that U.S. intelligence agencies were running mind-control experiments, infiltrating and disrupting civil-rights groups, or carrying out surreptitious missions at home like they do abroad had been fully internalized, and fused with the suspicion that Google, Facebook, Amazon, and Walmart were—in their own ways—following suit."

Buona lettura,

https://www.theatlantic.com/technology/archive/2025/04/american-panopticon/682616/

F.