What makes AI secure?

AI Basics

This episode details some of the AI ​​safety reports published in recent weeks. Jeffery Atik first covers the basics of attention-based AI, then he introduces reports on AI safety from OpenAI and Stanford. Exactly what AI safety covers remains unclear (and toxic in my view, after an ideological purge committed in the name of “trust and safety” by Silicon Valley’s content suppression bureaucracy). ). But there are arguably potential existential problems lurking beneath the surface of even the most ambitious AI projects.

Whether or not ChatGPT’s probabilistic parroting poses a threat to humanity, Nick Weaver reports that it clearly poses a threat to many people’s reputations.

I confess that the biggest data breaches of the last decade have surprisingly little cyber law. We know that a data leaker can do as much damage as a cyber spy just by folding up sensitive documents and taking pictures. While there is some evidence that the Russian government may have piggybacked on the leaks to sow the seeds of disinformation, Nick says the real mystery is the leakers’ motives. This leads to the question of whether being fur is a reason to lose clearance.

Paul Rosenzweig teaches us about RESTRICT laws that endanger information and communication technology. This allows the administration to restrict or ban him from TikTok. He highlights the most prominent arguments against the bill. The bill’s authors, Senators Mark Warner (D-VA) and Senator John Tune (R-SD), respond to this criticism, but they say that the president’s discretion is limited only by It appears to be only in the midst of a parliamentary frenzy. parliamentary action.

Nick is impressed by the law enforcement actions that shut down the Genesis Market, where credentials were widely sold to hackers. In this operation, the data that the FBI seizes will pay off for years to come.

I warn anyone who quits classified information jobs to work in the private sector. The Director of National Intelligence has issued a new directive calling for (sort of) caution if your new employer has ties to a foreign government. May violate federal law. While the directive has detailed provisions on how intelligence agencies should communicate new post-employment restrictions to current employees, it provides little guidance to intelligence graduates who have already moved into the private sector. Is not.

Nick is enthusiastic about the harsh tone the Treasury Department has taken in its report on illicit financial risks in decentralized finance.

Paul and I cover a Utah bill that requires teens to get parental approval to participate in social media sites. Having ridiculed the Republican Party and its congressional delegation for her 20 years in an attempt to control the impact of the Internet on their children, it seems to me that the parents in her class are starting to worry about their children. looks like If the idea of ​​checking the age of Internet users is approved by the United Kingdom, Utah and the New Yorker, those who oppose the proposal may have a tougher time than the 90’s.

And with a quick hit:

Download Episode 452 (mp3)

The Cyberlaw Podcast can be subscribed using iTunes, Google Play, Spotify, Pocket Casts, or an RSS feed. As always, the Cyberlaw Podcast welcomes your feedback.Be sure to participate @ Stewart Baker on Twitter. Send questions, comments, suggestions for topics or interviewees to CyberlawPodcast@steptoe.com. Remember: If your suggested guest is on the show, we’ll send you the long-awaited Cyberlaw Podcast mug! The views expressed in this podcast are those of the speakers and do not reflect the views of their institution, clients, friends, family, or pets.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *