The Cyberlaw Podcast: Are Stealth Quotas the Cure for AI Bias?

This week the Business Software Alliance issued a new report on AI bias. Jane Bambauer and I come to much the same conclusion: It is careful, well-written, and a policy catastrophe in the making. The main problem? It tries to turn one of the most divisive issues in American life into a problem to be solved by technology. Apparently because that has worked so well in areas like content suppression. In fact, I argue, the report will be seen by many, especially in the center and on the right, as an effort to impose proportional representation quotas by stealth in a host of places that have never been the objects of such policies before. Less controversial, but only a little, is the U.S. government’s attempt to make government data available for training more AI algorithms. Jane more or less persuades me that this effort too will end in tears or stasis.

In cheerier news, the good guys got a couple of surprising wins this week. While encryption and bitcoin have posed a lot of problems for law enforcement in recent years, the FBI has responded with imagination and elan, at least if we can judge by two stories from last week. First, Nick Weaver takes us through the laugh-out-loud facts behind a, government-run encrypted phone for criminals complete with influencers, invitation-only membership, and nosebleed pricing to cement the phone’s exclusive status. Jane Bambauer unpacks some of the surprisingly complicated legal questions raised by the FBI’s creativity.

Paul Rosenzweig lays out the much more obscure facts underlying the FBI’s recovery of much of the ransom paid by Colonial Pipeline. There’s no doubt that the government surprised everyone by coming up with the private key controlling the bitcoin account. We’d like to celebrate the ingenuity behind the accomplishment, but the how it pulled it off, probably because it hopes to do the same thing again and can’t if it blows the secret. FBI isn’t actually explaining.

The Biden administration is again taking a shaky and impromptu Trump policy and giving it a sober interagency foundation. This time it’s the TikTok and WeChat bans; these have been rescinded. But a new process has been put in place that could restore and even expand those bans in a matter of months. Paul and I disagree about whether the Biden administration will end up applying the Trump policy to TikTok or WeChat or to a much larger group of Chinese apps.

For comic relief, Nick regales us with Brian Krebs’s wacky story of the FSB’s weird and counterproductive attempt to secure communications to the FSB’s web site.

Jane and I review the latest paper by Bruce Schneier (and Henry Farrell) on how to address the impact of technology on American democracy. We are not persuaded by its suggestion that our partisan divide can best be healed by more understanding, civility, and aggressive prosecutions of Republicans.

Finally, everyone confesses to some confusion about the claim that the Trump Justice Department breached norms in its criminal discovery motions that turned up records relating to prominent Democratic congressmen and at least one Trump administration official.

Best bet: this flap will turn out to be less interesting the more we learn. But I renew my appeal, this time aimed at outraged Democrats, for more statutory guardrails and safeguards against partisan misuse of national security authorities. Because that’s what we’ll need if we want to keep those authorities on the books.

And more!

Download the 366th Episode (mp3)

You can subscribe to The Cyberlaw Podcast using iTunes, Google Play, Spotify, Pocket Casts, or our RSS feed. As always, The Cyberlaw Podcast is open to feedback. Be sure to engage with @stewartbaker on Twitter. Send your questions, comments, and suggestions for topics or interviewees to CyberlawPodcast@steptoe.com. Remember: If your suggested guest appears on the show, we will send you a highly coveted Cyberlaw Podcast mug!

The views expressed in this podcast are those of the speakers and do not reflect the opinions of their institutions, clients, friends, families, or pets.