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

The Last Watchdog
The Last Watchdog
The GitHub Blog
The GitHub Blog
博客园_首页
Engineering at Meta
Engineering at Meta
B
Blog
罗磊的独立博客
美团技术团队
人人都是产品经理
人人都是产品经理
云风的 BLOG
云风的 BLOG
L
LangChain Blog
腾讯CDC
Microsoft Azure Blog
Microsoft Azure Blog
H
Help Net Security
G
Google Developers Blog
Martin Fowler
Martin Fowler
Vercel News
Vercel News
大猫的无限游戏
大猫的无限游戏
D
Darknet – Hacking Tools, Hacker News & Cyber Security
P
Palo Alto Networks Blog
L
LINUX DO - 热门话题
T
Threatpost
IT之家
IT之家
P
Proofpoint News Feed
aimingoo的专栏
aimingoo的专栏
Recent Announcements
Recent Announcements
月光博客
月光博客
B
Blog RSS Feed
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
雷峰网
雷峰网
博客园 - 叶小钗
A
Arctic Wolf
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
Know Your Adversary
Know Your Adversary
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Google DeepMind News
Google DeepMind News
Cyberwarzone
Cyberwarzone
S
Schneier on Security
Cisco Talos Blog
Cisco Talos Blog
博客园 - 【当耐特】
V
Vulnerabilities – Threatpost
量子位
Project Zero
Project Zero
C
CXSECURITY Database RSS Feed - CXSecurity.com
博客园 - 聂微东
Help Net Security
Help Net Security
Google Online Security Blog
Google Online Security Blog
Security Latest
Security Latest
W
WeLiveSecurity
Simon Willison's Weblog
Simon Willison's Weblog
爱范儿
爱范儿

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
A New Tool for Curbing AI Cheating (guest post) - Daily Nous
Justin Weinberg · 2026-06-18 · via Hacker News - Newest: "AI"

“The aim is not to keep everything exactly as it was before gen AI took off. That would be both impossible and undesirable. The aim is to preserve the parts of philosophical education that are still worth preserving while changing the surrounding infrastructure enough to make that possible.”

That’s David Bourget, a philosophy professor at Western University and executive director of the PhilPapers Foundation, who has created a new tool to help professors—not just philosophy professors, but anyone who teaches by having students write—struggling with the prospect of their students using ChatGPT, Grammarly, and other AI tools to cheat on their assignments.

The tool, currently in beta testing, is called MATCHA (Modern Authoring Tool for Certified Human Authorship). You can try it out here. But before you do, you should read Professor Bourget’s post about it, below, where he describes its purpose, features, and limitations.


MATCHA: A New Tool for Curbing AI Cheating
by David Bourget

Over the past year or so, I’ve been working on a software project whose aim is to make cheating hard again. The result is an app called MATCHA, which stands for Modern Authoring Tool for Certified Human Authorship. In this post, I will explain how MATCHA differs from alternative ways of managing AI plagiarism, how it works, and what it can do beyond preventing AI plagiarism.

Current responses to AI plagiarism seem to fall into three main categories: detecting, scaffolding, and abandoning essay assignments. Detecting worked to some extent for a time, but it now seems both fraught and ineffective. Even analyzing detailed edit histories is largely ineffective on its own, because students can simply hand-transcribe AI-generated essays, giving them plausible deniability. Scaffolding an assignment into multiple steps does not solve the problem either, because ChatGPT can complete the individual steps just as easily as it can produce a finished essay. Abandoning essay assignments does get rid of plagiarism, but at the cost of abandoning a traditional pillar of liberal arts education.

There are two different approaches that MATCHA makes possible. The first is aimed at ordinary home assignments. The second is aimed at supervised writing labs. Both approaches are forms of prevention rather than detection. The goal is not to guess, after the fact, whether a text was produced by AI. The goal is to structure the writing process so that using AI is no longer the path of least resistance.

Start with home assignments, which is the case many of us care most about. MATCHA’s approach here has three key ingredients.

The first ingredient is the production of a detailed record of all work done within MATCHA—not just the editing history, as many Google Docs plug-ins and other tools provide, but a broader proof of work that includes browsing and reading activity, which takes place within MATCHA’s built-in, whitelist-controlled browser (MATCHA only monitors browsing within its own browser). When a student submits a project, the instructor receives a summary of how much of each cited or otherwise relevant work the student has read, along with the more standard editing-history statistics. In cases of doubt, the reading and writing process can be replayed. One available graphical summary of reading looks like this (it’s possible to view stats about specific works by clicking the slices):

This is not meant to be magic. A proof of work is not a proof of the entire inner process within a student’s mind. Students can think offline. They can read printed books. They can have conversations. Etc. But a plausible record of reading, drafting, revising, and engaging with sources is still evidence of something. And the absence of such a record can also be evidence of something, especially when producing the record was part of the assignment.

The second part is in-person authorship-assessment quizzes. The idea is simple. If a student really wrote the paper, they should usually be able to answer questions about it: why they made certain choices, what a cited author argues, how one part of the paper connects to another, what they would say in response to an obvious objection, and so on. MATCHA can help generate student-specific questions and provide an initial assessment of the answers, so that this does not create an unreasonable amount of extra work for instructors. But the instructor remains in control. The generated questions should be reviewed, and the final judgment should be the instructor’s.

The third ingredient is to make these things matter. Students need to know that they are expected to submit a plausible proof of work and to be able to answer questions about their own papers. This can be done in different ways, but the basic idea is to tie part of the grade to the writing record, the reading record, and the follow-up quiz.

Even when put together, these measures do not make AI cheating impossible. Nothing does. What they do is make successful AI cheating significantly harder. In many cases, it may become about as hard as doing the work properly. That is a large improvement over the current situation.

The second approach made possible by MATCHA is the supervised writing lab. I find this approach especially promising, though it requires more institutional coordination. MATCHA can block AI tools, other apps, and automation on students’ computers. If someone can supervise the writing to prevent other devices from being used, AI plagiarism can be prevented to roughly the same extent as other forms of cheating in a supervised writing environment. This gives us an alternative to the usual forced choice: we do not have to choose between ordinary take-home essays, where AI use may be very hard to control, and in-class essay exams, where students have too little time to do long-form work. A supervised writing lab allows students to write over time, but in a controlled environment. MATCHA supports this with a check-in/check-out system that allows supervisors to enable writing when students are in the space and block it when they leave.

There are obvious logistical challenges. Someone has to provide the space. Someone has to supervise it. At Western, we are fortunate to have some available rooms and TA hours that can be used for this purpose. A key point, though, is that the supervision need not be especially heavy-handed. In many cases, a TA can supervise a lab while doing other work, such as grading. This means that some regular TA hours can contribute to lab supervision.

MATCHA can also be used for other in-person activities. Its Single-App Mode blocks all other apps, not just AI tools, which makes it suitable for writing exams. It can also be used for in-class note-taking. If students take notes in MATCHA, this eliminates the usual laptop distractions without having to ban laptops altogether. Since MATCHA creates a centralized attendance record for each activity, instructors can also use it to assign participation credit.

I should say that MATCHA isn’t about avoiding AI entirely. The app has its own built-in AI assistant, which can help with writing and (soon) research. Crucially, however, instructors control the level of help available (for example, “grammar only”). All the work done by the assistant is tracked and labeled. The assistant also has a pedagogical mode, which makes it turn errors into practice exercises: instead of correcting every error, it highlights passages that need revision and provides hints, giving positive or negative feedback as the student tries to improve the text.

I should be explicit about the status of the project. MATCHA is not just a pedagogical proposal or a research prototype. It is software intended for real use, and it will eventually be a paid product, since operating and maintaining it involves real costs. At the moment, it is still in beta testing. Instructors who are interested in trying it with their students can request access, and beta users will not be charged for their first year of use. The beta period is partly for improving the app and partly for learning how it works in different teaching contexts.

Even if you are not teaching or don’t need something like this for your courses, you might want to check out MATCHA as a word processor. I gave it many features that I wanted in a word processor, such as the ability to insert citations directly from in-app PhilPapers searches, or interoperability with both LaTeX and Word. These and other features (live collaboration, automatic proxy browsing, etc.) can make the lives of academics much easier. All the anti-AI-cheating features can be turned off when MATCHA is used outside educational settings, so they don’t get in the way (but it might soon be valuable for some of us to be able to point at a MATCHA proof of work).

Postscript

When I talk about this project with a large enough group, someone usually raises a version of the following worry: isn’t this just conservatism? Why try to preserve old writing assignments? Why not adapt to AI in the same way that math instruction adapted to calculators?

The calculator analogy is mistaken. First, we do forbid calculators in early education, precisely so that students can develop basic arithmetic skills. If the analogy worked, it would suggest that we should restrict AI while students are still learning to write, argue, interpret texts, and respond to objections. Second, calculators never threatened to do all the relevant work for students. A calculator can carry out computations, but it does not do the whole work we are trying to assess in math courses. AI is different: it can do all the work we might reasonably ask of an undergraduate student.

I’m not opposed to AI in general. There are many good uses of AI, including in teaching, research, and elsewhere (people like to point out that MATCHA’s website has AI-generated illustrations; yes, it does). But there is a real danger in allowing AI to replace the very activities through which students are supposed to develop their abilities. In philosophy, those activities include reading difficult texts, reconstructing arguments, finding objections, revising one’s own claims, and writing with care. These are not incidental to philosophical education. They are a large part of it.

The aim is not to keep everything exactly as it was before gen AI took off. That would be both impossible and undesirable. The aim is to preserve the parts of philosophical education that are still worth preserving while changing the surrounding infrastructure enough to make that possible. I hope MATCHA can help bring long-form essay writing back to the center of the philosophy curriculum—not because the old ways were perfect, but because students still need the kind of training that writing assignments provide.