Peter Henderson
@peterhenderson.bsky.social
Assistant Professor the Polaris Lab @ Princeton (https://www.polarislab.org/); Researching: RL, Strategic Decision-Making+Exploration; AI+Law
We’ve been pushing hard on AI for public good. One example: partnering with Courtlistener to launch accessible legal semantic search! Many more cool AI projects coming soon from my group aimed at improving access to justice, often spearheaded by @dominsta.bsky.social !
November 7, 2025 at 2:15 AM
We’ve been pushing hard on AI for public good. One example: partnering with Courtlistener to launch accessible legal semantic search! Many more cool AI projects coming soon from my group aimed at improving access to justice, often spearheaded by @dominsta.bsky.social !
Sora2 is speedrunning my AI law class. We covered issues with copyrighted characters in week 2, and right of publicity claims in week 3. Georgia has a postmortem right of publicity claim. Some states don't (e.g., famous Marilyn Monroe estate battle).
October 17, 2025 at 8:06 PM
Sora2 is speedrunning my AI law class. We covered issues with copyrighted characters in week 2, and right of publicity claims in week 3. Georgia has a postmortem right of publicity claim. Some states don't (e.g., famous Marilyn Monroe estate battle).
October 16, 2025 at 9:14 PM
Quick take: Are open-weight AI models getting a fair shake in evals? A few thoughts on comparing systems-to-models, sparked by Anthropic’s recent postmortem.
Check it our most recent post: www.ailawpolicy.com/p/quick-take...
Check it our most recent post: www.ailawpolicy.com/p/quick-take...
September 24, 2025 at 3:15 PM
Quick take: Are open-weight AI models getting a fair shake in evals? A few thoughts on comparing systems-to-models, sparked by Anthropic’s recent postmortem.
Check it our most recent post: www.ailawpolicy.com/p/quick-take...
Check it our most recent post: www.ailawpolicy.com/p/quick-take...
GPT-5-codex just ``git reset --hard'' ongoing changes in a repo, saying "I panicked!"
h/t Zeyu Shen @ Princeton
h/t Zeyu Shen @ Princeton
September 23, 2025 at 6:34 PM
GPT-5-codex just ``git reset --hard'' ongoing changes in a repo, saying "I panicked!"
h/t Zeyu Shen @ Princeton
h/t Zeyu Shen @ Princeton
Annnnnndddd Judge Alsup just rejected the settlement. Still some time to fix it. Rejection was mostly on the grounds that the class was under-specified (no final list of works, no opt-out/notification mechanism solidified).
news.bloomberglaw.com/ip-law/anthr...
news.bloomberglaw.com/ip-law/anthr...
September 8, 2025 at 11:48 PM
Annnnnndddd Judge Alsup just rejected the settlement. Still some time to fix it. Rejection was mostly on the grounds that the class was under-specified (no final list of works, no opt-out/notification mechanism solidified).
news.bloomberglaw.com/ip-law/anthr...
news.bloomberglaw.com/ip-law/anthr...
The terms of Anthropic's settlement w/book authors just came out.
💰$1.5B to authors in libgen (Books3 corpus)!
Interestingly, this is ~$3k per book, close to the terms that HarperCollins allegedly gave to authors for their books ($2.5k). Consensus price forming?
💰$1.5B to authors in libgen (Books3 corpus)!
Interestingly, this is ~$3k per book, close to the terms that HarperCollins allegedly gave to authors for their books ($2.5k). Consensus price forming?
September 5, 2025 at 7:59 PM
The terms of Anthropic's settlement w/book authors just came out.
💰$1.5B to authors in libgen (Books3 corpus)!
Interestingly, this is ~$3k per book, close to the terms that HarperCollins allegedly gave to authors for their books ($2.5k). Consensus price forming?
💰$1.5B to authors in libgen (Books3 corpus)!
Interestingly, this is ~$3k per book, close to the terms that HarperCollins allegedly gave to authors for their books ($2.5k). Consensus price forming?
Wonder why Claude decided to report users to the authorities? It might be because its constitution says Claude should choose responses in the long-term interest of humanity!
But what if we could leverage computational and legal tools to "debug" or "lint" AI rules/laws for ambiguity?
🧵!
But what if we could leverage computational and legal tools to "debug" or "lint" AI rules/laws for ambiguity?
🧵!
September 5, 2025 at 1:57 PM
Wonder why Claude decided to report users to the authorities? It might be because its constitution says Claude should choose responses in the long-term interest of humanity!
But what if we could leverage computational and legal tools to "debug" or "lint" AI rules/laws for ambiguity?
🧵!
But what if we could leverage computational and legal tools to "debug" or "lint" AI rules/laws for ambiguity?
🧵!
Excited to offer my AI Law class again @ Princeton this year. We'll be sharing lecture notes/materials and more this year on the course webpage! Imo, we have a unique offering that emphasizes how the technical details affect legal outcomes. Check it out!
www.polarislab.org/ai-law-2025/...
www.polarislab.org/ai-law-2025/...
September 4, 2025 at 11:25 PM
Excited to offer my AI Law class again @ Princeton this year. We'll be sharing lecture notes/materials and more this year on the course webpage! Imo, we have a unique offering that emphasizes how the technical details affect legal outcomes. Check it out!
www.polarislab.org/ai-law-2025/...
www.polarislab.org/ai-law-2025/...
New paper suggests that if firms aren’t seeing growth from AI, it could be because current deployments replace existing labor, instead of scaling output. AI policy and governance agenda for 2025+ needs to put labor at the forefront.
digitaleconomy.stanford.edu/publications...
digitaleconomy.stanford.edu/publications...
August 26, 2025 at 2:30 PM
New paper suggests that if firms aren’t seeing growth from AI, it could be because current deployments replace existing labor, instead of scaling output. AI policy and governance agenda for 2025+ needs to put labor at the forefront.
digitaleconomy.stanford.edu/publications...
digitaleconomy.stanford.edu/publications...
AI-generated errors in an Australian murder case. We'll probably see an influx of ineffective assistance of counsel petitions/appeals soon arguing AI-usage.
apnews.com/article/aust...
apnews.com/article/aust...
August 21, 2025 at 1:46 PM
AI-generated errors in an Australian murder case. We'll probably see an influx of ineffective assistance of counsel petitions/appeals soon arguing AI-usage.
apnews.com/article/aust...
apnews.com/article/aust...
New work from Hartline, Hu & Wu: is there a truthful calibration metric in sequential settings (i.e., better than ECE)? Seems like the answer is yes! Super important research direction as we think about multi-step uncertainty estimation from agents in high stakes settings.
August 20, 2025 at 6:13 PM
New work from Hartline, Hu & Wu: is there a truthful calibration metric in sequential settings (i.e., better than ECE)? Seems like the answer is yes! Super important research direction as we think about multi-step uncertainty estimation from agents in high stakes settings.
August 4, 2025 at 5:21 PM
July 14, 2025 at 10:22 PM
Check out our new blogpost and policy brief on our recently updated lab website!
❓Are we actually capturing the bubble of risk for cybersecurity evals? Not really! Adversaries can modify agents by a small amount and get massive gains.
❓Are we actually capturing the bubble of risk for cybersecurity evals? Not really! Adversaries can modify agents by a small amount and get massive gains.
July 14, 2025 at 10:22 PM
Check out our new blogpost and policy brief on our recently updated lab website!
❓Are we actually capturing the bubble of risk for cybersecurity evals? Not really! Adversaries can modify agents by a small amount and get massive gains.
❓Are we actually capturing the bubble of risk for cybersecurity evals? Not really! Adversaries can modify agents by a small amount and get massive gains.
We're up 216 tracked cases of bogus citations in court worldwide, including this case!
www.polarislab.org/ai-law-track...
www.polarislab.org/ai-law-track...
July 4, 2025 at 10:27 PM
We're up 216 tracked cases of bogus citations in court worldwide, including this case!
www.polarislab.org/ai-law-track...
www.polarislab.org/ai-law-track...
First time a judge has decided a case based on hallucinated case law in the US that I've encountered.
caselaw.findlaw.com/court/ga-cou...
caselaw.findlaw.com/court/ga-cou...
July 3, 2025 at 11:35 PM
First time a judge has decided a case based on hallucinated case law in the US that I've encountered.
caselaw.findlaw.com/court/ga-cou...
caselaw.findlaw.com/court/ga-cou...
📉Lots more in this paper, but one quick point! If you head to the website, you'll also see trends. And some frontier models have *decreased* in Elo rating on newer problems since April 2025.
From April 2025 -> June 2025
Gemini 2.5 Pro 03-25: 1983 ->1282
o3-mini (01-31): 1896 → 1403
From April 2025 -> June 2025
Gemini 2.5 Pro 03-25: 1983 ->1282
o3-mini (01-31): 1896 → 1403
June 17, 2025 at 1:18 PM
📉Lots more in this paper, but one quick point! If you head to the website, you'll also see trends. And some frontier models have *decreased* in Elo rating on newer problems since April 2025.
From April 2025 -> June 2025
Gemini 2.5 Pro 03-25: 1983 ->1282
o3-mini (01-31): 1896 → 1403
From April 2025 -> June 2025
Gemini 2.5 Pro 03-25: 1983 ->1282
o3-mini (01-31): 1896 → 1403
Excited for this work to be out!
🔍Benchmarks, in my opinion, should break down performance to figure out what aggregate metrics might miss. Here, we showed the weaknesses of current frontier models for competition coding and where there's room to grow at a granular level!
livecodebenchpro.com
🔍Benchmarks, in my opinion, should break down performance to figure out what aggregate metrics might miss. Here, we showed the weaknesses of current frontier models for competition coding and where there's room to grow at a granular level!
livecodebenchpro.com
June 17, 2025 at 1:18 PM
Excited for this work to be out!
🔍Benchmarks, in my opinion, should break down performance to figure out what aggregate metrics might miss. Here, we showed the weaknesses of current frontier models for competition coding and where there's room to grow at a granular level!
livecodebenchpro.com
🔍Benchmarks, in my opinion, should break down performance to figure out what aggregate metrics might miss. Here, we showed the weaknesses of current frontier models for competition coding and where there's room to grow at a granular level!
livecodebenchpro.com
Thanks to the new utm_source field that ChatGPT attaches to links, we see a lot of filings in court that clearly show evidence of ChatGPT usage despite not yet being called out or do not contain hallucinated citations.
June 17, 2025 at 1:14 PM
Thanks to the new utm_source field that ChatGPT attaches to links, we see a lot of filings in court that clearly show evidence of ChatGPT usage despite not yet being called out or do not contain hallucinated citations.
Faiz Surani, one of our project team leads, built an improved interactive map for AB1466-covered Racially Restrictive Covenants in Santa Clara County—with extra granular detail on a subset of deeds from 1923–1939, including the actual discriminatory language. Check it out! 🔗👇
sccmap.reglabapp.com
sccmap.reglabapp.com
June 9, 2025 at 12:03 PM
Faiz Surani, one of our project team leads, built an improved interactive map for AB1466-covered Racially Restrictive Covenants in Santa Clara County—with extra granular detail on a subset of deeds from 1923–1939, including the actual discriminatory language. Check it out! 🔗👇
sccmap.reglabapp.com
sccmap.reglabapp.com
June 5, 2025 at 10:03 AM
🚨Reddit sues Anthropic!🚨
This is going to be a really interesting case. Some quick thoughts... 🧵👇
1️⃣ Notice: no copyright claim. Reddit doesn't really own the copyright to user-uploaded content, so nothing to do here. Reddit also doesn't make any federal claims to keep it in state court.
This is going to be a really interesting case. Some quick thoughts... 🧵👇
1️⃣ Notice: no copyright claim. Reddit doesn't really own the copyright to user-uploaded content, so nothing to do here. Reddit also doesn't make any federal claims to keep it in state court.
June 5, 2025 at 10:03 AM
🚨Reddit sues Anthropic!🚨
This is going to be a really interesting case. Some quick thoughts... 🧵👇
1️⃣ Notice: no copyright claim. Reddit doesn't really own the copyright to user-uploaded content, so nothing to do here. Reddit also doesn't make any federal claims to keep it in state court.
This is going to be a really interesting case. Some quick thoughts... 🧵👇
1️⃣ Notice: no copyright claim. Reddit doesn't really own the copyright to user-uploaded content, so nothing to do here. Reddit also doesn't make any federal claims to keep it in state court.
FWIW we’re now at 167 cases of nonexistent law/cases being cited across the world. It’s a mix of pro se litigants, attorneys, and even adjudicators. AI is here to stay. Even if attorneys stop using it, pro se litigants won’t.
June 4, 2025 at 9:19 AM
FWIW we’re now at 167 cases of nonexistent law/cases being cited across the world. It’s a mix of pro se litigants, attorneys, and even adjudicators. AI is here to stay. Even if attorneys stop using it, pro se litigants won’t.
Ccompute budgets are a useful measure to quantify the cost of an adversary growing the risk bubble. For example, our study finds that 8 GPU-hours grew an offensive cyber-agent’s success on InterCode-CTF by +40% using relatively simple methods.
June 3, 2025 at 4:16 PM
Ccompute budgets are a useful measure to quantify the cost of an adversary growing the risk bubble. For example, our study finds that 8 GPU-hours grew an offensive cyber-agent’s success on InterCode-CTF by +40% using relatively simple methods.