AI surveillance is moving fast, and governments seem A-OK with that

You'd think that all over the world, all sides would agree on the need to place strict limits on intrusive AI. After all, everyone, including those in power, is potentially vulnerable to it. But apparently you'd be wrong, at least for now.

By all accounts, the European Union’s AI Act seems like something that would make tech ethicists happy. The landmark artificial intelligence law, which enjoyed a huge legislative win this week, seeks to institute a broad regulatory framework that would tackle the harms posed by the new technology. In its current form, the law would go to some lengths to make AI more transparent, inclusive, and consumer-friendly in the countries where it’s in effect. What’s not to like, right?

...And yet, despite some policy victories, there are major blindspots in the new law—ones that make civil society groups more than a little alarmed...

Indeed, the European Parliament’s own press release on the landmark law admits that “narrow exceptions [exist] for the use of biometric identification systems (RBI)”—police mass surveillance tech, in other words—“in publicly accessible spaces for law enforcement purposes, subject to prior judicial authorisation and for strictly defined lists of crime.” As part of that exemption, the law allows police to use live facial recognition technology—a controversial tool that has been dubbed “Orwellian” for its ability to monitor and catalog members of the public—in cases where it’s used to prevent “a specific and present terrorist threat” or to ID or find someone who is suspected of a crime.

As you might expect, for groups like Amnesty, this seems like a pretty big blind spot. From critics’ perspective, there’s no telling how law enforcement’s use of these technologies could grow in the future. “National security exemptions—as we know from history—are often just a way for governments to implement quite expansive surveillance systems,” said Satija. - Gizmodo