The Volokh Conspiracy

Mostly law professors | Sometimes contrarian | Often libertarian | Always independent

ChatGPT Successfully Imitates a Talented Sociopath with Too Many Lawyers

Episode 434 of the Cyberlaw Podcast


It's been a news-heavy week, but we have the most fun in this episode with ChatGPT. Jane Bambauer, Richard Stiennon, and I pick over the astonishing number of use cases and misuse cases disclosed by the release of ChatGPT for public access. It is talented – writing dozens of term papers in seconds. It is sociopathic – the term papers are full of falsehoods, down to the made-up citations to plausible but nonexistent URLs for New York Times stories. And it has too many lawyers – Richard's request that it provide his bio (or even Albert Einstein's) was refused on what are almost certainly data protection grounds. Luckily, either ChatGPT or its lawyers are also bone stupid, since reframing the question tricks the machine into subverting the legal and PC limits it labors under. I speculate that it beat Google to a PR triumph precisely because Google had even more lawyers telling their Artificial Intelligence what not to say.

In a surprisingly undercovered story, Apple has stopped pretending to care about child pornography. Its phone encryption already makes the iPhone a safe place to record child sexual abuse material (CSAM); now Apple will encrypt users' cloud storage with keys it cannot access, allowing customers to upload CSAM without fear of law enforcement. And it has abandoned its effort to identify such material by doing phone-based screening. All that's left of its effort to stop such abuse is a feature allowing parents to force their kids to activate an option that prevents them from sending or receiving nude photos. Jane and I dig into the story, as well as Apple's questionable claim to be offering the same encryption to its Chinese customers.

Nate Jones brings us up to date on the National Defense Authorization Act, or NDAA. Lots of second-tier cyber provisions made it into the bill, but not the provision requiring that critical infrastructure companies report security breaches. A contested provision on spyware purchases by the U.S. government was compromised into a more useful requirement that the intelligence community identify spyware that poses risks to the government.

Jane updates us on what European data protectionists have in store for Meta, and it's not pretty. The EU data protection supervisory board intends to tell the Meta companies that they cannot give people a free social media network in exchange for watching what they do on the network and serving ads based on their behavior. If so, it's a one-two punch. Apple delivered the first blow by curtailing Meta's access to third-party behavioral data. Now even first-party data could be off limits in Europe. That's a big revenue hit, and it raises questions whether Facebook will want to keep giving away its services in Europe.  

Mike Masnick is Glenn Greenwald with a tech bent – often wrong but never in doubt, and contemptuous of anyone who disagrees. But when he's right, he's right. Jane and I discuss his article recognizing that data protection is becoming a tool that the rich and powerful can use to squash annoying journalist-investigators. I have been saying this for decades. But still, welcome to the party, Mike!

Nate points to a post pleading for more controls on the export of personal data from the U.S. It comes not from the usual privacy enthusiasts but from the U.S. Naval Institute, and it makes sense.

Jane and I take time to marvel at the story of France's Mr. Privacy and the endless appetite of Europe's bureaucrats for serial grifting, as long as it combines enthusiasm for American technology with hostility to the technology's source.

Nate and I cover what could be a good resolution to the snake-bitten cloud contract competition at the Department of Defense. The Pentagon is going to let four cloud companies—Google, Amazon, Oracle And Microsoft – share the prize.

You didn't think we'd forget Twitter, did you? Jane, Richard, and I all comment on the Twitter Files. Consensus: the journalists claiming these stories are nothingburgers are driven more by ideology than their nose for news. Especially newsworthy are the remarkable proliferation of shadowbanning tools Twitter developed for suppressing speech it didn't like, and some considerable though anecdotal evidence that Twitter's many speech rules were often twisted to suppress speech from the right—even when the rules did not quite fit, as with LibsofTikTok—while similar behavior on the left went unpunished. Richard tells us what it feels like to be on the receiving end of a Twitter shadowban. 

The podcast introduces a new feature: "We Read It So You Don't Have To," and Nate provides the tl;dr on an New York Times story: How the Global Spyware Industry Spiraled Out of Control.

And in quick hits and updates:

Download the 434th 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 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.

Update: Corrected a typo and toned down criticism of Apple that the commenters persuaded me was unfair.