The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
If Biden Knocked Tiktok, Can Trump Deep-Six DeepSeek?
The government would have a much stronger interest to ban a Chinese AI app that collects proprietary information than to ban a social media app that displays dance videos.
Long-time readers know I am ambivalent about AI. I refuse to use it, even as I know others think it gives them a competitive advantage. I am confident in my ability to write, research, and ideate better without AI. Maybe I'm wrong, but so be it.
I also reacted with some ambivalence to the launch of DeepSeek, a new Chinese AI App. I have no clue whether it is better or worse than OpenAI. But I do worry about the fact that millions of Americans are installing a Chinese AI app on their phones, and will submit troves of proprietary information about anything on their minds. Remember, every query can be tracked by DeepSeek, and by extension, the PRC. We learned as much from the TikTok litigation. We aren't simply talking about people watching dance videos. People ask their AI apps about everything. And I suspect submitting privileged information to a foreign adversary waives any sort of Attorney-Client privilege. The risk of potential blackmail and other compromise is unlimited.
Moreover, as could be expected, the app is generating Pro-PRC answers. You can call it data manipulation. The New York Times reports: "If you're among the millions of people who have downloaded DeepSeek, the free new chatbot from China powered by artificial intelligence, know this: The answers it gives you will largely reflect the worldview of the Chinese Communist Party."
Whatever interest may have supported the TikTok ban seems even greater for DeepSeek. And while the TikTok creators arguably had a First Amendment right to use the app to make videos, there is not a similar expressive interest here. To be sure, I've long argued that some algorithmic outputs may warrant First Amendment scrutiny. But as we move towards general AI, which has very little intercession of the human mind, the First Amendment rights are harder to justify. (I anticipated this shift in a 2013 article.)
Biden knocked TikTok. Will Trump deep-six DeepSeek?
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
"...Long-time readers know I am ambivalent about AI. I refuse to use it, even as I know others think it gives them a competitive advantage...."
Shoot. My last hope for Josh was that he WAS using AI, and had just programmed the algorithm to churn out breathless fringe posts here. You mean, all of his endless posts really came from his own brain?!???
That's a scary thought. 🙁
Josh's posts are so predictable and formulaic that I just assumed he was an AI chaptbot.
Getting away from politics for the moment, I would recommend you go to YouTube and look up The Small Faces song "Watcha Gonna do about it"
> But as we move towards general AI, which has very little intercession of the human mind, the First Amendment right are harder to justify.
BUT THAT'S THE POINT OF General AI!
Are you saying an AGI entity would have no First Amendment rights!? This is why the ASI will do away with us!
If Biden Knocked Tiktok, Can Trump Deep-Six DeepSeek?
Do you think your legal arguments have the best chance of landing when you preface them with inane wordplay titles? I might be more receptive to your analysis if the first thing I saw didn't make my eyes roll.
That's the least objectionable thing about Blackman's posts, and there's a long tradition in law review articles of giving them cutesy wordplay titles.
On your first point I agree. I was trying to dial back my contempt. As for the wordplay, sure, even Eugene does it from time to time. But the operative words there are "from time to time," not every frigging time. When you read the titles in the Latest box, do you ever doubt you can identify Josh's with close to 100% accuracy from the clownish titles alone?
Is there a legal aspect to this blog post?
If you want to know what Trump will do, find a way to understand how he personally benefits in some way--that's what he will do.
DeekSeek "open sourced" their code. It's not really open source when you take a look at it, but code is speech. This is settled law. I forget the case, but DVD decryption code was ruled as speech years ago. You can post code and it is protected as 1A speech. Can't regulate that.
So he can try to deep six DeepSeek, they are in a slightly better position than TikTok, but not very. It brings the constitutional 1A issues to the forefront potentially in DeepSeek's case.
Can the president issue a blanket ban on things from China solely due to it being labeled as a "national security issue"?
https://en.wikipedia.org/wiki/DVD_Copy_Control_Ass%27n,_Inc._v._Bunner
He is not objecting to Deepseek posting its weights. He is objecting to Chinese controlled servers running the model for American users.
Blackman refuses to be associated in any way with anything involving the word "intelligence."
The ship has sailed. The cat is out of the bag. The die is cast.
Or, insert your preferred adage or idiom.
Established algorithms don't cease to exist by decree or rule. Those choosing to be blind to them do so at their own peril.
ChatGPT: How may I help you?
Josh: SHells Tasmanian articel
ChatGPT: Can you rephrase that?
Josh: Proof necklace Talisman evil
ChatGPT: Those shells are just shells
Josh: I refuse to use AI
Biden didn't do anything to TikTok besides sign a law of Congress.