返回新闻
政策AI Understanding 简报

印度最高法院撤销对人工智能生成判例法的 42.5 亿卢比罚款

Juris Hour 报道称,印度最高法院在发现该命令依赖于显然由人工智能生成的不存在或不准确代表的法律权威后,撤销了约 42.528 亿卢比的海关罚款。

4 min readRead the linked source
Source-provided image accompanying Indian Supreme Court sets aside ₹425 crore penalty over AI-generated case laws
来源参考来源记录
出版商
jurishour.in
来源链接
jurishour.inhttps://www.jurishour.in/indirect-taxes/customs-fake-ai-generated-case-laws/
来源类型
链接来源——主要来源状态尚未确定。
背景60 秒内了解这一点

从这里开始

关键术语

人工智能(AI)
构建执行需要模式识别、推理、语言或决策的任务的系统的广泛领域。
分类
模型将输入分配给一个或多个预定义类别的任务。
生成式 AI
生成文本、图像、音频、视频或代码等新内容的人工智能系统。
测试一下自己人工智能道德测验

发生了什么

Juris Hour reports that the Supreme Court of India set aside a customs penalty of ₹425,27,99,100 and a Gujarat High Court order after finding that the underlying customs decision cited nonexistent judgments, fake citations and legal propositions not supported by the decisions named. The case was remanded for fresh adjudication by an officer of the same rank, but not the officer who issued the original order.

Juris Hour reports that the dispute concerned an October 8, 2025 Order-in-Original issued by the Additional Commissioner of Customs in Surat. The customs department had imposed a penalty of ₹425,27,99,100 under Section 114 of the Customs Act, 1962, alleging that natural diamonds had been declared as lab-grown diamonds to obtain a lower tariff. The Gujarat High Court dismissed the appellant’s challenge on January 20, 2026, after which the matter reached the Supreme Court.

According to Juris Hour, the appellant argued that judgments and articles cited in the customs order had been generated using artificial intelligence. The Supreme Court reportedly checked the cited material itself and found that some decisions did not exist, some citations were fake, and some real decisions did not support the propositions attributed to them. The report says the Court described the resulting legal conclusions as apparently reflecting an AI hallucination.

Juris Hour reports that the Supreme Court set aside both the Gujarat High Court’s order and the customs authority’s original penalty order. The Court did not decide whether the diamonds had been misdeclared. Instead, it revived the customs proceedings and directed that they be heard again by an officer of the same rank as the original adjudicator, excluding the officer who issued the defective order.

The report says the Bench referred to an earlier Supreme Court decision, Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., and reiterated a zero-tolerance approach to unverified AI-generated precedents. It also cited draft 2026 regulations on AI use in courts, which had been released for comments but were not yet final according to the source.

来源详情: jurishour.in ↗

为什么这很重要

The reported ruling places human verification at the center of AI-assisted legal work. It shows that AI-generated citations can create consequences far beyond an incorrect answer: they may undermine a government penalty and require an entire proceeding to be reconsidered. The report also signals a strict judicial standard for using AI in adjudication, while leaving important details of the full order unavailable without Juris Hour membership.

The reported decision treats fabricated or misrepresented authorities as a fundamental integrity problem, not merely a minor drafting error. Juris Hour says the Court considered reliance on such material fatal to the sustainability of the penalty, even without resolving whether the false authorities directly changed the outcome. That approach raises the operational standard for lawyers, agencies and courts using : every citation and legal proposition must be checked against an authoritative source.

The case is practically relevant to any organization using AI for legal research, compliance or administrative decisions. A human reviewer needs to verify that cited cases exist, that citations are accurate, and that the cited holdings actually support the proposition being made. The report does not independently establish the full reasoning because the complete case details and order copy were restricted to Juris Hour members.

The source reports that the Court accepts AI as an assistive tool but rejects delegating adjudicatory responsibility to it. The practical implication is that AI may help locate or organize legal material, but responsibility for the final legal analysis remains with the human decision-maker.

Interactive Mechanism

互动机制:它实际上是如何运作的

以交互方式探索这一发展背后的基础技术。

Thinking Budget (Test-Time Tokens):1,024 tokens
Complex Accuracy79%Math & Code Logic
Latency3.2sTime to first full output
Inference Cost$0.0092Per query estimated
Reasoning StyleStep VerificationInternal chain depth
Active Thinking Trace:
1Deconstruct user problem into formal constraints
2Propose candidate hypotheses & step-by-step calculation
3Self-correction: Backtrack and refute subtle edge cases
4Exhaustive consistency check & final output synthesis
Core takeaway: Test-time compute fundamentally changes AI economics. Instead of only scaling during pre-training, giving reasoning models more tokens at inference time allows them to systematically solve PhD-level STEM problems.
交互式概念检查+10 Points
AI Ethics Quiz

Why can ethical evaluation not be reduced to one model score?

接下来看什么

Watch for the complete Supreme Court order, the fresh customs adjudication, and the final version of the Supreme Court’s proposed AI regulations. The source does not establish whether the alleged diamond misdeclaration occurred, whether the original officer faced any separate consequences, or how broadly the ruling will be applied beyond the reported case.

The complete Supreme Court judgment or order should clarify the precise authorities that were fabricated or misrepresented, the procedural basis for the remand, and whether the Court issued any further directions concerning the officer or the department. The source does not provide independent access to that document.

The fresh customs proceeding will determine the underlying allegation concerning the of the diamonds. The Supreme Court’s reported decision did not resolve that issue, so the setting aside of the penalty should not be treated as a finding that the alleged misdeclaration did not occur.

The Supreme Court’s draft Regulations for Use of Artificial Intelligence in Courts may establish more detailed requirements for disclosure, verification, supervision or record-keeping. Juris Hour reports that the regulations were still being considered and had not been finalized.

The source names an Additional Solicitor General who did not dispute interference and remand, but it does not report any separate disciplinary, administrative or professional action. Such consequences remain unknown.

相关指南和测验

AI 伦理ChatGPT 与大语言模型人工智能模型解释测试你所知道的——尝试免费的人工智能测验在我们的词汇表中查找人工智能术语关注AI监管追踪器
觉得这有用吗?