Josh Goldstein
@joshagoldstein.bsky.social
Research Fellow @Georgetown's Center for Security and Emerging Technology (CSET).
Studying emerging tech, national security, and online manipulation. Trying to bridge the academia/policy divide.
https://cset.georgetown.edu/staff/josh-a-goldstein/
Studying emerging tech, national security, and online manipulation. Trying to bridge the academia/policy divide.
https://cset.georgetown.edu/staff/josh-a-goldstein/
Lots going on about AI policy right now (+ lots more to come).
To help make sense of it all, my colleagues at CSET do deep-dives into AI workforce development, the impact of AI on cyber, US-China competition, military applications, bio risk, and more.
Check em out below:
bsky.app/starter-pack...
To help make sense of it all, my colleagues at CSET do deep-dives into AI workforce development, the impact of AI on cyber, US-China competition, military applications, bio risk, and more.
Check em out below:
bsky.app/starter-pack...
January 22, 2025 at 3:44 PM
Lots going on about AI policy right now (+ lots more to come).
To help make sense of it all, my colleagues at CSET do deep-dives into AI workforce development, the impact of AI on cyber, US-China competition, military applications, bio risk, and more.
Check em out below:
bsky.app/starter-pack...
To help make sense of it all, my colleagues at CSET do deep-dives into AI workforce development, the impact of AI on cyber, US-China competition, military applications, bio risk, and more.
Check em out below:
bsky.app/starter-pack...
Reposted by Josh Goldstein
In a new publication, @samlai.bsky.social, @benimmo.bsky.social, Derek Ruths and @lageneralista.ca lay out four key questions that underpin successful measurement of changes caused by GenAI:
Measuring Changes Caused by Generative Artificial Intelligence: Setting the Foundations
Informed policy that leads to beneficial change is extremely challenging to develop without being able to measure the material impacts of GenAI.
carnegieendowment.org
January 9, 2025 at 3:04 PM
In a new publication, @samlai.bsky.social, @benimmo.bsky.social, Derek Ruths and @lageneralista.ca lay out four key questions that underpin successful measurement of changes caused by GenAI:
Reposted by Josh Goldstein
This Friday (Jan 3rd) is the deadline to apply to be in the 2nd cohort of @goodauth.bsky.social fellows: goodauthority.org/news/goodaut...
Please share with scholars who want to bring insights from political science to a broader audience.
Please share with scholars who want to bring insights from political science to a broader audience.
Good Authority fellows program now open for applications!
Join the team at Good Authority! Our 2025-2026 fellow program is now accepting applications.
goodauthority.org
December 31, 2024 at 10:49 PM
This Friday (Jan 3rd) is the deadline to apply to be in the 2nd cohort of @goodauth.bsky.social fellows: goodauthority.org/news/goodaut...
Please share with scholars who want to bring insights from political science to a broader audience.
Please share with scholars who want to bring insights from political science to a broader audience.
The piece also includes two ideas for relatively low-cost, but perhaps meaningful, ways to expand reporting.
Curious for any reactions from T&S practitioners about feasibility.
Thanks to @justinhendrix.bsky.social for the quick edit. First time writing for TPP and would recommend to others!
Curious for any reactions from T&S practitioners about feasibility.
Thanks to @justinhendrix.bsky.social for the quick edit. First time writing for TPP and would recommend to others!
December 17, 2024 at 4:32 PM
The piece also includes two ideas for relatively low-cost, but perhaps meaningful, ways to expand reporting.
Curious for any reactions from T&S practitioners about feasibility.
Thanks to @justinhendrix.bsky.social for the quick edit. First time writing for TPP and would recommend to others!
Curious for any reactions from T&S practitioners about feasibility.
Thanks to @justinhendrix.bsky.social for the quick edit. First time writing for TPP and would recommend to others!
Wrote a piece in @techpolicypress.bsky.social on some of the lessons that Meta’s Q3 threat report brings back to the surface in online IO:
1. Tension between virality & opsec
2. Experimentation is ongoing
3. Variation in defenses creates arbitrage dynamics
4. Line btwn overt & covert is blurry
1. Tension between virality & opsec
2. Experimentation is ongoing
3. Variation in defenses creates arbitrage dynamics
4. Line btwn overt & covert is blurry
December 17, 2024 at 4:32 PM
Wrote a piece in @techpolicypress.bsky.social on some of the lessons that Meta’s Q3 threat report brings back to the surface in online IO:
1. Tension between virality & opsec
2. Experimentation is ongoing
3. Variation in defenses creates arbitrage dynamics
4. Line btwn overt & covert is blurry
1. Tension between virality & opsec
2. Experimentation is ongoing
3. Variation in defenses creates arbitrage dynamics
4. Line btwn overt & covert is blurry
Reposted by Josh Goldstein
🚀🏛️🚀 Introducing AGORA, ETO's new explorer for the AI governance landscape - with plain-English summaries, full text, and thematic tags for AI laws and policies around the world, from D.C. to Sacramento to Beijing 🌎 🗺️ agora.eto.tech
December 11, 2024 at 4:07 PM
🚀🏛️🚀 Introducing AGORA, ETO's new explorer for the AI governance landscape - with plain-English summaries, full text, and thematic tags for AI laws and policies around the world, from D.C. to Sacramento to Beijing 🌎 🗺️ agora.eto.tech
One @aaron.bsky.team line that struck me on Bluesky's labeling features:
"It's premature to say the experiment has succeeded or failed when we are so rapidly growing everything at the same time."
"It's premature to say the experiment has succeeded or failed when we are so rapidly growing everything at the same time."
December 5, 2024 at 6:31 PM
One @aaron.bsky.team line that struck me on Bluesky's labeling features:
"It's premature to say the experiment has succeeded or failed when we are so rapidly growing everything at the same time."
"It's premature to say the experiment has succeeded or failed when we are so rapidly growing everything at the same time."
What's it like to build platform policies in real time, as new waves of users pour in?
@aaron.bsky.team pulls back layers on @katieharbath.bsky.social's podcast with quite interesting examples about the challenges that have come on the radar at Bluesky (copyright, spam, youth verification, etc.)
@aaron.bsky.team pulls back layers on @katieharbath.bsky.social's podcast with quite interesting examples about the challenges that have come on the radar at Bluesky (copyright, spam, youth verification, etc.)
Bluesky's VP of Trust and Safety Aaron Rodericks @aaron.bsky.team joins me on the Impossible Tradeoffs podcast to discuss how he and the company are managing the surge of users since the U.S. election. 🎧 Listen Now! anchorchange.substack.com/p/blueskys-a...
Bluesky's Approach to Trust and Safety
Aaron Rodericks talks through the tradeoffs the decentralized platform is facing with rapid growth
anchorchange.substack.com
December 5, 2024 at 6:31 PM
What's it like to build platform policies in real time, as new waves of users pour in?
@aaron.bsky.team pulls back layers on @katieharbath.bsky.social's podcast with quite interesting examples about the challenges that have come on the radar at Bluesky (copyright, spam, youth verification, etc.)
@aaron.bsky.team pulls back layers on @katieharbath.bsky.social's podcast with quite interesting examples about the challenges that have come on the radar at Bluesky (copyright, spam, youth verification, etc.)
From a misuse perspective (i.e., disinfo/influence ops)
Persuasiveness compared to content from real campaigns:
academic.oup.com/pnasnexus/ar...
A field study by Kreps/Kriner on forging constituent mail:
journals.sagepub.com/doi/abs/10.1...
Linvill/Warren and their team have recent cases studies
Persuasiveness compared to content from real campaigns:
academic.oup.com/pnasnexus/ar...
A field study by Kreps/Kriner on forging constituent mail:
journals.sagepub.com/doi/abs/10.1...
Linvill/Warren and their team have recent cases studies
How persuasive is AI-generated propaganda?
Abstract. Can large language models, a form of artificial intelligence (AI), generate persuasive propaganda? We conducted a preregistered survey experiment
academic.oup.com
December 5, 2024 at 12:04 PM
From a misuse perspective (i.e., disinfo/influence ops)
Persuasiveness compared to content from real campaigns:
academic.oup.com/pnasnexus/ar...
A field study by Kreps/Kriner on forging constituent mail:
journals.sagepub.com/doi/abs/10.1...
Linvill/Warren and their team have recent cases studies
Persuasiveness compared to content from real campaigns:
academic.oup.com/pnasnexus/ar...
A field study by Kreps/Kriner on forging constituent mail:
journals.sagepub.com/doi/abs/10.1...
Linvill/Warren and their team have recent cases studies
Comments based on my work with @noupside.bsky.social in the HKS Misinformation Review on the use of AI-generated images by spammers/scammers/creators for audience growth.
misinforeview.hks.harvard.edu/article/how-...
misinforeview.hks.harvard.edu/article/how-...
How spammers and scammers leverage AI-generated images on Facebook for audience growth | HKS Misinformation Review
Much of the research and discourse on risks from artificial intelligence (AI) image generators, such as DALL-E and Midjourney, has centered around whether they could be used to inject false informatio...
misinforeview.hks.harvard.edu
December 3, 2024 at 9:29 PM
Comments based on my work with @noupside.bsky.social in the HKS Misinformation Review on the use of AI-generated images by spammers/scammers/creators for audience growth.
misinforeview.hks.harvard.edu/article/how-...
misinforeview.hks.harvard.edu/article/how-...
Oxford dictionary's word of the year: "brain rot."
@shiraovide.bsky.social prefers another one from the shortlist: "slop."
“Spammers are opportunists,” I shared.
“They’re constantly asking: What are people online interested in, and what technologies are available?”
@shiraovide.bsky.social prefers another one from the shortlist: "slop."
“Spammers are opportunists,” I shared.
“They’re constantly asking: What are people online interested in, and what technologies are available?”
Sorry, Oxford dictionary nerds. This is the real word of the year.
The verbal encapsulation of 2024 explains everything from “shrimp Jesus” to disgusting viral recipes and clickbait headlines.
wapo.st
December 3, 2024 at 9:29 PM
Oxford dictionary's word of the year: "brain rot."
@shiraovide.bsky.social prefers another one from the shortlist: "slop."
“Spammers are opportunists,” I shared.
“They’re constantly asking: What are people online interested in, and what technologies are available?”
@shiraovide.bsky.social prefers another one from the shortlist: "slop."
“Spammers are opportunists,” I shared.
“They’re constantly asking: What are people online interested in, and what technologies are available?”
I’ve been waiting (/hoping?) for this one! 👏👏
December 3, 2024 at 9:04 PM
I’ve been waiting (/hoping?) for this one! 👏👏
Reposted by Josh Goldstein
Digital identity is going to be a hugely important space in the next few years. A lot of complaints in this thread highlight the reasons why.
There's a whole lot of conversation still to be had about proving identity, but just want to call out that this option is available right now (and super cool to see!)
The only way Dylan could get a nytimes-dot-com handle like this is if nytimes set it up!
The only way Dylan could get a nytimes-dot-com handle like this is if nytimes set it up!
Update: I'm now @dylanfreedman.nytimes.com!
December 2, 2024 at 11:31 PM
Digital identity is going to be a hugely important space in the next few years. A lot of complaints in this thread highlight the reasons why.
Haven’t seen much in terms of quantitative assessments of impact, but UT Austin put together a set of international case studies: mediaengagement.org/research/gen...
Generative Artificial Intelligence and Elections
The Center for Media Engagement investigates GenAI’s role before, during, and after several key global elections in 2024.
mediaengagement.org
November 30, 2024 at 4:22 AM
Haven’t seen much in terms of quantitative assessments of impact, but UT Austin put together a set of international case studies: mediaengagement.org/research/gen...
Reposted by Josh Goldstein
Starter pack of ppl working in influence operations, disinformation, propaganda, information environment space. This is an iteration of a list I maintained in the bad place called “disinformation braintrust.” I’m selective here to avoid the bs artists. Lemme know if I missed ppl
go.bsky.app/4URqqgz
go.bsky.app/4URqqgz
November 16, 2024 at 1:05 AM
Starter pack of ppl working in influence operations, disinformation, propaganda, information environment space. This is an iteration of a list I maintained in the bad place called “disinformation braintrust.” I’m selective here to avoid the bs artists. Lemme know if I missed ppl
go.bsky.app/4URqqgz
go.bsky.app/4URqqgz
Thanks @justinhendrix.bsky.social for coming to @cset.bsky.social today to chat with the analysis team about all things pitching & publishing in the tech and democracy space.
We’ll be following @techpolicypress.bsky.social in the year ahead!
We’ll be following @techpolicypress.bsky.social in the year ahead!
November 22, 2024 at 12:16 AM
Thanks @justinhendrix.bsky.social for coming to @cset.bsky.social today to chat with the analysis team about all things pitching & publishing in the tech and democracy space.
We’ll be following @techpolicypress.bsky.social in the year ahead!
We’ll be following @techpolicypress.bsky.social in the year ahead!
Just read your testimony and it’s a great overview of a bunch of different threat vectors.
Planning to add to my course syllabus for the spring. Thanks!
Planning to add to my course syllabus for the spring. Thanks!
November 20, 2024 at 11:35 PM
Just read your testimony and it’s a great overview of a bunch of different threat vectors.
Planning to add to my course syllabus for the spring. Thanks!
Planning to add to my course syllabus for the spring. Thanks!
Free for a coffee during the day on the 27th?
I'll be back in NY for a few days & would be keen to chat about silicon sampling in social science (opportunities, risks) in the vein of your post earlier. I have an ongoing collaborative project in that area as well.
I'll be back in NY for a few days & would be keen to chat about silicon sampling in social science (opportunities, risks) in the vein of your post earlier. I have an ongoing collaborative project in that area as well.
November 20, 2024 at 10:57 PM
Free for a coffee during the day on the 27th?
I'll be back in NY for a few days & would be keen to chat about silicon sampling in social science (opportunities, risks) in the vein of your post earlier. I have an ongoing collaborative project in that area as well.
I'll be back in NY for a few days & would be keen to chat about silicon sampling in social science (opportunities, risks) in the vein of your post earlier. I have an ongoing collaborative project in that area as well.