Jonathan Mayer
@mayer.bsky.social
Princeton prof and tech + law person. Previously at the Justice Department, Senate, FCC, Stanford, and CalDOJ. Views are solely my own.
The full text of the AI Executive Order is now available on the White House website. Complete with, as @brianfung.me notes, a fun Halloween bat GIF. www.whitehouse.gov/briefing-roo...
Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence ...
By the authority vested in me as President by the Constitution and the laws of the United States of America, it is hereby ordered as follows: Section 1. Purpose. Artificial intelligence (A...
www.whitehouse.gov
October 30, 2023 at 9:06 PM
The full text of the AI Executive Order is now available on the White House website. Complete with, as @brianfung.me notes, a fun Halloween bat GIF. www.whitehouse.gov/briefing-roo...
The Jamaal Bowman fire alarm thing is a great example of a cheapfake. Presumably the video shows him struggling to open the door to get to the House floor. But the Administration Committee posted a selective still photo, generating a news cycle about attempting to delay the vote to end the shutdown.
October 1, 2023 at 12:55 PM
The Jamaal Bowman fire alarm thing is a great example of a cheapfake. Presumably the video shows him struggling to open the door to get to the House floor. But the Administration Committee posted a selective still photo, generating a news cycle about attempting to delay the vote to end the shutdown.
Could we not do the FTC Amazon complaint, FCC net neutrality NPRM, and DOJ Google trial testimony on the same day? Some of us are trying to get tech law research done over here.
September 26, 2023 at 6:13 PM
Could we not do the FTC Amazon complaint, FCC net neutrality NPRM, and DOJ Google trial testimony on the same day? Some of us are trying to get tech law research done over here.
What if the Kansas newspaper raid was legal? And what if that entirely depended on… the user interface design of a website?
As a criminal procedure and computer crime person, looking at the undisputed public facts so far, what the police did may have been lawful (but awful). Allow me to explain…
As a criminal procedure and computer crime person, looking at the undisputed public facts so far, what the police did may have been lawful (but awful). Allow me to explain…
August 14, 2023 at 2:26 AM
What if the Kansas newspaper raid was legal? And what if that entirely depended on… the user interface design of a website?
As a criminal procedure and computer crime person, looking at the undisputed public facts so far, what the police did may have been lawful (but awful). Allow me to explain…
As a criminal procedure and computer crime person, looking at the undisputed public facts so far, what the police did may have been lawful (but awful). Allow me to explain…
Here’s a great example of why survey design is so important. NPR/Marist asked about raising the debt ceiling so the government can “pay its bills” & “avoid a default.” 52-42 for a clean increase. CNN/SSRS gave a wordy & budget-ish prompt (“keep all government programs running”). 60-24 the other way!
May 24, 2023 at 12:51 PM
Here’s a great example of why survey design is so important. NPR/Marist asked about raising the debt ceiling so the government can “pay its bills” & “avoid a default.” 52-42 for a clean increase. CNN/SSRS gave a wordy & budget-ish prompt (“keep all government programs running”). 60-24 the other way!
OpenAI’s grand AI policy proposal is… a new international agency focused on long-term risks of AI that exceeds human cognition. It’s like they’re straight up trolling the AI fairness, governance, etc. communities, which consistently emphasize the need for near-term actionable solutions to AI harms.
Governance of superintelligence
Now is a good time to start thinking about the governance of superintelligence—future AI systems dramatically more capable than even AGI.
openai.com
May 22, 2023 at 9:13 PM
OpenAI’s grand AI policy proposal is… a new international agency focused on long-term risks of AI that exceeds human cognition. It’s like they’re straight up trolling the AI fairness, governance, etc. communities, which consistently emphasize the need for near-term actionable solutions to AI harms.
While Twitter's slapdash encrypted DMs are already old news, I want to highlight a drawback that’s sorta flown under the radar: one of the goals is integration with the Twitter website. That's a bad idea. You can’t build E2EE with ordinary web tech and achieve strong security and privacy guarantees.
May 12, 2023 at 8:15 PM
While Twitter's slapdash encrypted DMs are already old news, I want to highlight a drawback that’s sorta flown under the radar: one of the goals is integration with the Twitter website. That's a bad idea. You can’t build E2EE with ordinary web tech and achieve strong security and privacy guarantees.
I told my students that they could use a notebook or a tablet on the final exam. Now some are asking for clarification, whether it would be OK to instead use a laptop. I feel so old.
May 12, 2023 at 1:43 PM
I told my students that they could use a notebook or a tablet on the final exam. Now some are asking for clarification, whether it would be OK to instead use a laptop. I feel so old.
In our introductory undergrad coursework on information security, we cover: (1) getting encrypted messaging right is far more complicated than public-private key pairs and session keys, and 2) never ever roll your own crypto. We literally teach the students not to do exactly what Twitter is doing.
May 10, 2023 at 11:51 PM
In our introductory undergrad coursework on information security, we cover: (1) getting encrypted messaging right is far more complicated than public-private key pairs and session keys, and 2) never ever roll your own crypto. We literally teach the students not to do exactly what Twitter is doing.
Hot take: what if Bluesky didn’t add DMs, but rather, acted as a directory for Signal or other E2EE messaging apps? It would be easy to associate keys with usernames/DIDs, without revealing phone or email info. Apps could implement message permissions with filtering (e.g., only people you follow).
May 9, 2023 at 1:43 PM
Hot take: what if Bluesky didn’t add DMs, but rather, acted as a directory for Signal or other E2EE messaging apps? It would be easy to associate keys with usernames/DIDs, without revealing phone or email info. Apps could implement message permissions with filtering (e.g., only people you follow).
On the upside, it’s now easier to explain online survey sampling bias in one screenshot.
May 8, 2023 at 7:03 PM
On the upside, it’s now easier to explain online survey sampling bias in one screenshot.
Twitter’s paid “verification” just got worse, again. The blue check used to have a popup with a quasi-disclaimer that verification meant a subscription and a working phone. That text is gone.
On the upside, the argument that Twitter “verification” violates consumer deception law just got stronger.
On the upside, the argument that Twitter “verification” violates consumer deception law just got stronger.
May 3, 2023 at 3:54 PM
Twitter’s paid “verification” just got worse, again. The blue check used to have a popup with a quasi-disclaimer that verification meant a subscription and a working phone. That text is gone.
On the upside, the argument that Twitter “verification” violates consumer deception law just got stronger.
On the upside, the argument that Twitter “verification” violates consumer deception law just got stronger.
The 6-month LLM pause didn’t stick, but could we do a 6-month LLM clout chasing pause on here? It’s rather nice to not doomscroll through endless “one weird trick with ChatGPT,” “here’s a trivial test of GPT-3.5 vs. GPT-4,” “watch me dunk on LLM developers,” and all the other genre favorites.
April 30, 2023 at 5:43 PM
The 6-month LLM pause didn’t stick, but could we do a 6-month LLM clout chasing pause on here? It’s rather nice to not doomscroll through endless “one weird trick with ChatGPT,” “here’s a trivial test of GPT-3.5 vs. GPT-4,” “watch me dunk on LLM developers,” and all the other genre favorites.