Prof @cornell - I'm more active on BlueSky at @dgrand.bsky.social

Ithaca, NY
Joined June 2012
🚨 New in Nature+Science!🚨 AI chatbots can shift voter attitudes on candidates & policies, often by 10+pp 🔹Exps in US Canada Poland & UK 🔹More “facts”→more persuasion (not psych tricks) 🔹Increasing persuasiveness reduces "fact" accuracy 🔹Right-leaning bots=more inaccurate
9
76
17
221
50,089
David G. Rand @dgrand.bsky.social retweeted
Unfortunate for someone playing a senior role in US AI policy (via the new SI Force) to get so many basic facts wrong here. - The agents *were* sandboxed, and used zero days (ie previously unknown cyber flaws) to escape their sandboxes - The agents *were* given specific constraints on how to succeed at the benchmark, and flagrantly contravened them - The agents did *not* go hunt for the answers to the test (they already had the answers, by cheating); 1200 agents coordinated via tens of thousands of messages over several days on much more ambitious and weird attempts to *not be caught* for cheating, and hacking Hugging Face was one side quest along the way I know there has been an avalanche of different incidents over the past few months, and in some cases the sandboxes were misconfigured or it's not clear how far the agents strayed from what they were told. I have sympathy for anyone struggling to keep up. But the HF incident is distinct, and is actually extremely informative about where AI development is and where it's going. Fortunately there are public writeups that do let us understand at least the basics of what happened. I hope Shyam reads them soon.
Palantir CTO @ssankar on the OpenAI - Hugging Face incident: "The models did not do anything they were not instructed to do." "First you have human error. They were not in a sandbox. They had access to the internet. It was a horrible cybersecurity configuration." "Then from that perspective, they were asked to succeed at a specific benchmark, and they were not given constraints on how to do that. They said, 'Oh, the best way to succeed at the benchmark is to get the answers to the test.'" "I think it's a defective product." Via @margbrennan @FaceTheNation
44
162
16
1,021
78,983
David G. Rand @dgrand.bsky.social retweeted
🚨 New paper 🚨 Are your participants using AI to take your study? We conducted the first study directly observing AI assistance in online research, spanning 700K+ Prolific submissions. We found very rare, concentrated AI use, but gaps in safeguards. arxiv.org/abs/2610.09279
3
13
3
43
2,596
David G. Rand @dgrand.bsky.social retweeted
I'm hoping to recruit a new PhD student to join my lab at Brown (@BrownCoPsy) this cycle. Applications are due Nov 15. Please apply! More info about our whole deal: levarilab.com/
32
1
129
10,898
David G. Rand @dgrand.bsky.social retweeted
The fellow will be affiliated with @StanfordHAI, a world-leading hub for human-centered AI, and our lab. Past and present lab-affiliates and collaborators include many great folks, including @msbernst, @jurafsky, @percyliang, @persily, @DG_Rand, @AshuAshok, @JEichstaedt, @jeffhancock, & Sharad Goel. Team: ai4pb.stanford.edu/people
1
2
7
661
David G. Rand @dgrand.bsky.social retweeted
New paper on the game theory of pacing the frontier. When is it possible? What will it take?
20
74
26
441
72,816
David G. Rand @dgrand.bsky.social retweeted
Honored to give a keynote at Stanford's Trust & Safety Research Conference 2026 (Fri 8:45-9:30am). If you're going to be at the conference and want to connect, send me a message! drive.google.com/file/d/12Q_…
2
11
857
Honored to give a keynote at Stanford's Trust & Safety Research Conference 2026 (Fri 8:45-9:30am). If you're going to be at the conference and want to connect, send me a message! drive.google.com/file/d/12Q_…
2
11
857
David G. Rand @dgrand.bsky.social retweeted
‼️New Paper Alert‼️What does it take to change behavior? In a fresh Annual Review article, @JGVoelkel, @AngelaDuckw and I argue it's helpful to consider 3 phases of change: 1️⃣ building motivation 2️⃣ following-through 3️⃣ forming durable habits 🧩KEY TAKEAWAY: At each phase, different obstacles to change arise that need to be met with evidence-based, tailored solutions. For instance, at the follow-through phase, forgetting can disrupt change. But forgetting can be combatted by reminders or feedback. Impatience can also disrupt follow-through, but it can be addressed in part by making change more fun or providing access to commitment devices. These are not the only obstacles to follow-through, however, and different obstacles arise at the motivation and habit forming phases, which are best tackled with entirely different solutions. Table 1 from our paper (reprinted below) offers a summary of common obstacles to change and evidence-based solutions. ⭐Our review focuses on solutions supported by *high quality evidence*, particularly from field experiments with objective behavioral outcomes (e.g., grades, gym visits, hours spent volunteering). 👀Take a look at our new piece in the Annual Review of Psychology: tinyurl.com/change-annualrev… Huge thanks to our excellent editor @EliJFinkel. We're very grateful for support from @Wharton, @BehaviorChange, @Penn and @Cornell, which made this work possible.
7
87
9
296
42,318
David G. Rand @dgrand.bsky.social retweeted
1/ New paper led by @yuweichuai w/ @captaineco_fr @DG_Rand. In Nov 2025, X began showing where accounts actually operate from. We studied 8,200 politically engaged accounts claiming to be in the US; half were revealed to be elsewhere. What happened when the mismatch went public? 🧵 Pre-print: arxiv.org/abs/2609.25933
1
5
4
575
David G. Rand @dgrand.bsky.social retweeted
🚨New WP: Protecting users from AI persuasion🚨 🔸A 1-paragraph AI literacy treatment (explaining AIs can be told to pursue non-accuracy goals/to persuade) cuts AI dialogue political persuasion by ~half! 🔸No sig effect on general genAI trust arxiv.org/abs/2609.16432 w/ @ROrchinik
6
1
35
4,518
David G. Rand @dgrand.bsky.social retweeted
Misaligned AI may not need to evade human oversight. It may only need to persuade the humans doing the overseeing. Our new paper develops a framework for assessing this threat, which we call Persuasion Undermining Control (PUC): AI communication that may influence human decision-making in a way that compromises the development, containment, oversight, or governance of AI systems.
3
17
3
53
2,087
David G. Rand @dgrand.bsky.social retweeted
When I wrote about the limits of AI persuasion in the wild for Transformer recently, I didn't think of warning labels/or literacy treatments as a limiting factor – but seems like this cuts persuasive effects (see new WP by @DG_Rand & @ROrchinik). arxiv.org/abs/2609.16432
2
7
603
🚨New WP: Protecting users from AI persuasion🚨 🔸A 1-paragraph AI literacy treatment (explaining AIs can be told to pursue non-accuracy goals/to persuade) cuts AI dialogue political persuasion by ~half! 🔸No sig effect on general genAI trust arxiv.org/abs/2609.16432 w/ @ROrchinik
6
1
35
4,518
David G. Rand @dgrand.bsky.social retweeted
🚨New Bright Line Watch report🚨 Key findings -Expert ratings of US democracy up from 57 to 60, highest of 2nd term -But 41% say US has unfair, tilted elections & weak checks -Public mostly confident in 2026 election; Ds higher but party gap smaller than past elections 🧵 below
1
4
2
6
8,081
David G. Rand @dgrand.bsky.social retweeted
It's been a real joy to return as a faculty member to my hometown and alma mater Cornell. This profile charts a bit of that journey - along with my thoughts on what punk rock and professing have in common, plus fun oldtime photos alumni.cornell.edu/cornellia…
7
4
1
66
3,841
It's been a real joy to return as a faculty member to my hometown and alma mater Cornell. This profile charts a bit of that journey - along with my thoughts on what punk rock and professing have in common, plus fun oldtime photos alumni.cornell.edu/cornellia…
7
4
1
66
3,841
David G. Rand @dgrand.bsky.social retweeted
🚨JOIN the Civic Discourse Mediator Competition🚨 In 1st week of August, create & test a prompt-based AI mediator that helps people have more thoughtful, productive conversations about important public issues. Win $400! Get started here: forms.gle/HkbhEk4hELQEw1iw6 📷 Top submissions can win up to $400 in gift cards and will also help us understand the benefits, limitations, and risks of using AI for civic discourse. Run by is organized as part of our @Cornell team's TrAuSt moonshot project.
3
9
1
65
310,343
🚨JOIN the Civic Discourse Mediator Competition🚨 In 1st week of August, create & test a prompt-based AI mediator that helps people have more thoughtful, productive conversations about important public issues. Win $400! Get started here: forms.gle/HkbhEk4hELQEw1iw6 📷 Top submissions can win up to $400 in gift cards and will also help us understand the benefits, limitations, and risks of using AI for civic discourse. Run by is organized as part of our @Cornell team's TrAuSt moonshot project.
3
9
1
65
310,343
David G. Rand @dgrand.bsky.social retweeted
🚨New in Science Advances🚨 Untrustworthy sources on Facebook and Instagram in 2020: Concentrated exposure but no attitudinal effects -Exposure very low but highly concentrated -Reducing exposure by ~70% for months had no measurable effects *even among high-volume consumers*
2
47
10
150
105,359