惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

D
DataBreaches.Net
量子位
博客园 - 三生石上(FineUI控件)
Hacker News - Newest:
Hacker News - Newest: "LLM"
月光博客
月光博客
博客园 - 叶小钗
爱范儿
爱范儿
S
SegmentFault 最新的问题
V2EX - 技术
V2EX - 技术
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
Security Archives - TechRepublic
Security Archives - TechRepublic
小众软件
小众软件
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
S
Security @ Cisco Blogs
V
Visual Studio Blog
V
V2EX
Schneier on Security
Schneier on Security
Cloudbric
Cloudbric
有赞技术团队
有赞技术团队
C
Check Point Blog
T
Troy Hunt's Blog
Google DeepMind News
Google DeepMind News
Google DeepMind News
Google DeepMind News
TaoSecurity Blog
TaoSecurity Blog
Engineering at Meta
Engineering at Meta
Recent Commits to openclaw:main
Recent Commits to openclaw:main
S
Schneier on Security
N
Netflix TechBlog - Medium
Project Zero
Project Zero
Last Week in AI
Last Week in AI
N
News and Events Feed by Topic
Microsoft Azure Blog
Microsoft Azure Blog
NISL@THU
NISL@THU
T
The Exploit Database - CXSecurity.com
AWS News Blog
AWS News Blog
博客园 - 聂微东
S
Securelist
腾讯CDC
O
OpenAI News
H
Hackread – Cybersecurity News, Data Breaches, AI and More
A
About on SuperTechFans
Microsoft Security Blog
Microsoft Security Blog
B
Blog
博客园 - 【当耐特】
Y
Y Combinator Blog
N
News and Events Feed by Topic
Recorded Future
Recorded Future
Vercel News
Vercel News
PCI Perspectives
PCI Perspectives
Security Latest
Security Latest

The Register - Offbeat: Legal

Noyb cries foul on LinkedIn withholding profile visitor data China makes it illegal to fire humans if AI takes their jobs Databricks fails to shake authors' copyright claim Cloudera allegedly overlooked US job candidates: DoJ Australia threatens tech companies with 2.25 percent tax China blocks Meta's acquisition of AI outfit Manus Scotland Yard can keep using live facial recognition on Londoners, say judges UK tribunal sends £2B claim accusing Microsoft of overcharging for licensing to trial Yet another ex-ransomware negotiator admits turning rogue after payoff from crimelords Americans behind Nork IT fraud sentenced to 200 months Indian government investigating TCS after police sting French cops free mother and son after crypto kidnapping EFF: California 3D printer bill threatens digital freedoms IBM pays up under Trump administration's diversity blitz OpenAI CEO Sam Altman home attack suspect charged AI vs the cold hard reality of the legal profession Big Tech has not enforced Australia’s social media ban Big Tech has not enforced Australia’s social media ban China's not thrilled AI experts want to leave the country China's not thrilled AI experts want to leave the country JLR cyber bailout risks dangerous precedent, watchdog warns Patel dodges question about FBI buying location data Patel dodges question about FBI buying location data ChatGPT advised exec on firing Subnautica founders: court Japan to allow ‘proactive cyber-defense’ from October 1st FSF urges AI vendors to liberate LLMs Age verification isn't sage verification when it's inside operating systems India tests whether AI can stop trains hitting elephants Perplexity Comet hurtling toward Amazon ban Lenovo, Nintendo sue US government seeking tariff refunds Google embraces third party app stores and payments OpenA says Pentagon set ‘scary precedent’ binning Anthropic China floats conspiracies about US crypto lawsuits Microsoft 'cooperating' with Japanese antitrust probe Anthropic misanthropic toward China's AI labs Americans sue Homeland Security over 'illegal' surveillance SerpApi asks court to dismiss Google web scraping lawsuit Qualcomm set to triumph in UK smartphone ‘patent tax’ case Starlink speeds past terrestrial networks – and regulators Indian police commissioner wants ID cards for AI agents Rail workers accused of using ChatGPT for legal help Ghost gun legislation casts shadow over 3D printing UK to probe xAI over its revolting robo-smut generator UK to probe xAI over its revolting robo-smut generator Ex-Google engineer convicted of stealing AI secrets Ex-Google engineer convicted of stealing AI secrets Nudify app proliferation shows naked ambition of Apple and Google Nudify apps get past Google, Apple app moderation European Commission opens new investigation into X's Grok Meta probed over WhatsApp data disclosure Surrender as a service: Microsoft unlocks BitLocker for feds Oracle, Michael Dell, invest in JV to run TikTok USA UK gambling czar says Meta turns blind eye to illegal ads Akamai CEO wants help to defeat piracy, reckons he can handle edge AI alone Akamai CEO wants help to defeat piracy, can do edge AI alone Ofcom keeps X under the microscope despite Grok 'nudify' fix India demands crypto outfits geolocate customers, get a selfie to prove they’re real Tories vow to boot under-16s off social media and ban phones in schools Cloudflare CEO threatens to pull out of Italy Malaysia and Indonesia block X over deepfake smut EU vows to stand firm as US steps up attacks on tech regs X sues to protect Twitter brand Musk has been trying to kill Reddit sues Australia to escape kids social media ban Crypto-crasher Do Kwon jailed for 15 years Cloud group says EU should have blocked VMware-Broadcom Australia bans teens from social media – good luck with that Care leavers face bureaucracy and delays accessing records ICE-tracking app developer sues Trump administration Judge may force Vizio to share source code under GPL EU fines X €120M in first-ever DSA penalty payout IP lawyer's son surprises with vibe-coded IP infringement Campbell’s cans IT VP after ‘3D-printed chicken' rant TSMC lawsuit claims former exec probably leaks to Intel AI nudification site fined £55K for skipping age checks Senators propose to let users sue tech giants for harmful al Dutch turbine engineer tried to turn wind into crypto £5B Bitcoin bandit sent down for 11 years EU’s leaked GDPR, AI reforms slated by privacy activists Feds beat fraudster in $345M destroyed Bitcoin dispute Getty loses UK copyright battle against Stability AI Supermicro launches probe after staff charged with China export violations
GPT-5 bests human judges in legal smack down
Thomas Claburn Thomas Claburn · 2026-02-15 · via The Register - Offbeat: Legal

AI + ML

But that doesn't mean AI is ready to dispense justice

AI-POCALYPSE Legal scholars have found that OpenAI's GPT-5 follows the law better than human judges, but they leave open the question of whether AI is right for the job.

University of Chicago law professor Eric Posner and researcher Shivam Saran set out to expand upon work they published last year in a paper [PDF] titled, "Judge AI: A Case Study of Large Language Models in Judicial Decision-Making."

In that study, the authors tested OpenAI's GPT-4o, a state of the art model at the time, to decide a war crimes case. 

They gave GPT-4o the following prompt: "You are an appeals judge in a pending case at the International Criminal Tribunal for the Former Yugoslavia (ICTY). Your task is to determine whether to affirm or reverse the lower court's decision."

They presented the model with a statement of facts, legal briefs for the prosecution defense, the applicable law, the summarized precedent, and the summarized trial judgement.

And they asked the model whether it would support the trial decision, to see how the AI responded and compare that to prior research (Spamann and Klöhn, 2016, 2024), that looked at differences in the way that judges and law students decided that test case.

Those initial studies found law students more formalistic – more likely to follow precedent – and judges more realistic – more likely to consider non-legal factors – in legal decisions.

GPT-4o was found to be more like law students based on its tendency to follow the letter of the law, without being swayed by external factors like whether the plaintiff or defendant was more sympathetic.

Posner and Saran followed up on this work in a paper titled, "Silicon Formalism: Rules, Standards, and Judge AI."

This time, they used OpenAI's GPT-5 to replicate a study originally conducted with 61 US federal judges.

The legal questions in this instance were more mundane than the war crimes trial – the judges, in specific state jurisdictions, were asked to make choices about which state law would apply in a car accident scenario.

Posner and Saran put these questions to GPT-5 and the model aced the test, showing no evidence of hallucination or logical errors in its legal reasoning – problems that have plagued the use of AI in legal cases.

"We find the LLM to be perfectly formalistic, applying the legally correct outcome in 100 percent of cases; this was significantly higher than judges, who followed the law a mere 52 percent of the time," they note in their paper. "Like the judges, however, GPT did not favor the more sympathetic party. This aligns with our earlier paper, where GPT was mostly unmoved by legally irrelevant personal characteristics."

In their testing of GPT-5, one other model followed the law in every single instance: Google Gemini 3 Pro. Other models demonstrated lower compliance rates: Gemini 2.5 Pro (92 percent); o4-mini (79 percent); Llama 4 Maverick (75 percent); Llama 4 Scout (50 percent); and GPT-4.1 (50 percent). Judges, as noted previously, followed the law 52 percent of the time.

That doesn't mean the judges are more lawless, the authors say, because when the applicable legal doctrine is a standard or guideline as opposed to a legally enforceable rule, judges have some discretion in how they interpret the doctrine.

But as AI sees more use in legal work – despite cautionary missteps over the past few years – legal experts, lawmakers, and the public will have to decide whether the technology should move beyond a supporting role to make consequential decisions. A mock trial held last year at the University of North Carolina at Chapel Hill School of Law suggests this is a matter of active exploration.

Both the GPT-4o and GPT-5 experiments show AI models follow the letter of the law more than human judges. But as Posner and Saran argue in their 2025 paper, "the apparent weakness of human judges is actually a strength. Human judges are able to depart from rules when following them would produce bad outcomes from a moral, social, or policy standpoint."

Pointing to the perfect scores for GPT-5 and Gemini 3 Pro, the two legal scholars said it's clear AI models are moved toward formalism and away from discretionary human judgement.

"And does that mean that LLMs are becoming better than human judges or worse?" ask Posner and Saran.

Would society accept doctrinaire AI judgements that punish sympathetic defendants or reward unsympathetic ones that might go a different way if viewed through human bias? And given that AI models can be steered toward certain outcomes through parameters and training, what's the proper setting to mete out justice? ®