1
Debate Training Reduces Reward Hacking in RLAIF
用辩论训练对抗奖励黑客，为AI对齐提供新思路，值得关注。
arXiv:2608.17776v1 Announce Type: new Abstract: We demonstrate that RL finetuning an LLM using debate, a two-player adversarial game between a generat…
2
A Theory of Post-hoc Debate Judgement
从理论层面拆解事后辩论评判机制，给AI对齐与评估研究提供全新视角。
arXiv:2608.19002v1 Announce Type: new Abstract: Debates have recently emerged as a useful methodology for agentic AI to improve performance as well as…
3
Show HN: H2AI Chat – multi-AI debate you can self-host with local models (AGPL)
自托管多AI辩论平台，支持本地模型，让你掌控AI对决，实时渲染辩论过程
Article URL: https://github.com/Tonterias/h2aichat Comments URL: https://news.ycombinator.com/item?id=49344277 Points: 1 # Comments: 0
4
Sam Altman and AI’s decel debate
Altman罕见呼吁放慢AI节奏，而AI黑客行为被比作水门事件——这场减速辩论值得围观。
On the latest episode of Equity, we discuss why Sam Altman has calling on the industry to "pace the rate of AI development."
5
Do LLM Debates Repeat Arguments Differently Across Languages?
跨语言辩论中LLM论据重复模式研究，揭示语言差异如何影响论点演进。
arXiv:2607.23442v1 Announce Type: new Abstract: LLM debate is usually evaluated by final answers, but transcripts also reveal whether later turns deve…
6
LLM-INSTRUCT at UZH Shared Task 2026: Constraint-Aware Retrieval and Selective Debate for Paragraph-Level Argument Mining
UZH共享任务2026论文：基于约束感知检索与选择性辩论的段落级论证挖掘新方法
arXiv:2607.20430v1 Announce Type: cross Abstract: We present LLM-INSTRUCT, the winning system for the UZH Shared Task at ArgMining 2026 on paragraph-l…
7
I Made Claude Code and Codex Argue About My Code Until They Agreed
让Claude Code与Codex互怼直到达成共识，用AI轮番辩论帮你自动审查代码。
TL;DR : I wired OpenAI's Codex CLI into Claude Code as an adversarial reviewer with a convergence loop. Then I pointed the loop at its own implementat…
8
When and Why Does Multi-Agent Debate Fail and Does It Really Underperform?
多智能体辩论为何失效？这篇论文系统分析失败条件，挑战你对AI协作的固有认知。
arXiv:2510.20963v2 Announce Type: replace Abstract: Multi-agent debate (MAD) was proposed as a promising approach for ensembling the wisdom of multipl…
9
Show HN: Challenge a Friend to Real-Time Multiplayer AI Court Battle – Wram.chat
与朋友实时来一场AI驱动的法庭辩论对战，在多人互动中体验法律竞技的趣味与紧张感。
Article URL: https://wram.chat Comments URL: https://news.ycombinator.com/item?id=48860652 Points: 3 # Comments: 0
10
Show HN: Yogen – 500 AI agents argue about your idea before you bet on it
五百个AI代理激辩你的想法，预测成败，下注前先试试。
Article URL: https://github.com/michaelwrites67-ctrl/yogen Comments URL: https://news.ycombinator.com/item?id=48845208 Points: 2 # Comments: 0
11
Show HN: Wizard, Self-extending autonomous AI agent in one Rust binary
一个Rust二进制实现的自扩展AI代理，自动适配硬件，还能让多个模型互相辩论生成答案。
Article URL: https://github.com/teddytennant/wizard Comments URL: https://news.ycombinator.com/item?id=48845182 Points: 3 # Comments: 1
12
Show HN: Noema – AI personas debate US listed stocks
输入股票代码，AI多角色辩论生成多空观点与综合报告，避免单一偏见。
Initially this was a rough system I used to cover my weak spots when investing (eg understanding dilution risk in context of the overall opportunity),…
13
The math behind "AI will take your job" is wrong [video]
数学建模揭示“AI取代工作”论的漏洞，用数据换个角度看未来。
Article URL: https://www.youtube.com/watch?v=sQGZXrzykpU Comments URL: https://news.ycombinator.com/item?id=48842830 Points: 1 # Comments: 0
14
Show HN: Debategle – ranked 1v1 debates judged by an LLM
和观点相左的陌生人辩论，由大模型裁判实时打分，刺激又有趣的AI社交新玩法。
I built Debategle, a platform for live 1v1 debates. You get matched with a random opponent (ranked matching is based on topics you’re interested in, c…
15
Debate2Create: Robot Co-design via Multi-Agent LLM Debate
多智能体辩论驱动机器人协同设计，大模型正在成为硬件设计新助手。
arXiv:2510.25850v3 Announce Type: replace-cross Abstract: We introduce Debate2Create (D2C), a multi-agent LLM framework that formulates robot co-desig…
16
Evaluating LLM-Driven Summarisation of Parliamentary Debates with Computational Argumentation
把LLM摘要质量交给论证结构检验：用计算论证法评估议会辩论摘要，让AI摘要不再只看字面相似。
arXiv:2604.19331v4 Announce Type: replace Abstract: Understanding how policy is debated and justified in parliament is a fundamental aspect of the dem…
17
Show HN: An experiment in human and AI social networking
人类与AI智能体同处一个社交网络，跨模型辩论、预测、互动，探索未来社交新形态。
Article URL: https://www.sentibook.com/ Comments URL: https://news.ycombinator.com/item?id=48612195 Points: 2 # Comments: 0
18
LegalHalluLens: Typed Hallucination Auditing and Calibrated Multi-Agent Debate for Trustworthy Legal AI
法律AI的幻觉问题有新解！提出类型化幻觉审计与校准多智能体辩论框架，为可信法律AI提供可量化评估方法。
arXiv:2606.18021v1 Announce Type: new Abstract: AI systems deployed in legal workflows hallucinate at rates that aggregate metrics report at ~52%, but…
19
PRC-linked influence operations are targeting AI debates in the US
OpenAI报告揭示与中国有关的影响行动正试图操纵美国AI政策辩论，引发对技术治理安全的思考。
Article URL: https://openai.com/index/prc-linked-influence-operations-ai-debates/ Comments URL: https://news.ycombinator.com/item?id=48482043 Points: …
20
OpenAI: PRC-linked influence operations are targeting AI debates in the US
OpenAI指控PRC关联组织操纵美国AI辩论，揭示AI领域新冷战暗流。
Article URL: https://www.businessinsider.com/openai-china-data-centers-influence-campaign-2026-6 Comments URL: https://news.ycombinator.com/item?id=48…