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

Microsoft Security Blog
Microsoft Security Blog
Jina AI
Jina AI
量子位
博客园 - 叶小钗
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
IT之家
IT之家
S
SegmentFault 最新的问题
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
小众软件
小众软件
Hugging Face - Blog
Hugging Face - Blog
雷峰网
雷峰网
博客园 - 聂微东
美团技术团队
Last Week in AI
Last Week in AI
罗磊的独立博客
酷 壳 – CoolShell
酷 壳 – CoolShell
博客园 - 三生石上(FineUI控件)
WordPress大学
WordPress大学
宝玉的分享
宝玉的分享
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
博客园_首页
V
Visual Studio Blog
大猫的无限游戏
大猫的无限游戏
The Cloudflare Blog

Hacker News

GitHub - SeanFDZ/macmind: Single-layer transformer in HyperTalk for the classic Macintosh Show HN: Agent-cache – Multi-tier LLM/tool/session caching for Valkey and Redis Bonsai 1-bit WebGPU - a Hugging Face Space by webml-community Moving a large-scale metrics pipeline from StatsD to OpenTelemetry / Prometheus GitHub - Nightmare-Eclipse/RedSun: The Red Sun vulnerability repository GitHub - SethPyle376/hiraeth: Local AWS emulator focused on fast integration testing, with SQS support, SQLite-backed state, and a debug-friendly web UI. GitHub - macOS26/Agent: Any AI, replaces Claude Code, Cursor, OpenClaw. Over 18 LLM providers (Claude, OpenAI, Gemini, Ollama, Zai, HF, Qwen) wired into a native Mac app that writes code, builds Xcode projects, bumps versions, manages git, automates Safari, use AppleScript, JS or Accessibility, extend Agent! w/ MCP Servers, run tasks from your iPhone via Messages. YouTube now lets you turn off Shorts I Made a Terminal Pager Burgers | マクドナルド公式 Commands — HackerNews CLI documentation ChatGPT for Excel PiCore - Raspberry Pi Port of Tiny Core Linux Live Nation illegally monopolized ticketing market, jury finds Google Broke Its Promise to Me. Now ICE Has My Data. Founding Engineer at Adaptional | Y Combinator CRISPR takes important step toward silencing Down syndrome’s extra chromosome GitHub - saffron-health/libretto: The AI toolkit for building reliable browser automations US v. Heppner (S.D.N.Y. 2026) no attorney-client privilege for AI chats [pdf] Retrofitting JIT Compilers into C Interpreters IPv6 – Google The Accursèd Alphabetical Clock Cybersecurity Looks Like Proof of Work Now Fragments: April 14 Cal.com Goes Closed Source: Why AI Security Is Forcing Our Decision | Cal.com - Scheduling Software for Online Bookings Laravel raised money and now injects ads directly into your agent When moving fast, talking is the first thing to break Too much Discussion of the XOR swap trick – Heather Cafe Introduction to Spherical Harmonics for Graphics Programmers The Grand Line
MAREF — GitHub Is Becoming a Giant AI Code Dump
MAREF Engineering · 2026-06-24 · via Hacker News

GitHub has 630 million repos. Nearly half of all new code is written by AI.

Sounds like a productivity explosion, right?

But there's another number in that same report: developer trust in AI code dropped from 77% to 60%. More people are using it. Fewer people believe in it.

And it gets worse.

Half the projects you see are fake

Carnegie Mellon University found 6 million fake stars on GitHub. Security firm Socket uncovered 370,000 "fix stars" directly tied to scams. That vibe coding project with thousands of stars? Probably half are fabricated.

That's not even the scary part.

CodeRabbit scanned 470 PRs. AI-written code had 1.7x more critical issues than human-written code. 45% of AI code ships with OWASP Top 10 vulnerabilities. 63% of developers say fixing AI code takes longer than writing it from scratch.

But here's the most mind-bending finding.

You're getting slower, but you have no idea

Meter ran a randomized controlled trial. The result: AI users were actually 19% slower, but they thought they were 20% faster.

When they showed the participants the data, they still insisted they were faster.

That's the real horror. You're getting slower, and you have no idea.

"This is metacognitive dissonance — the tool gives you the illusion of speed, but reality gives you a slap."

Open source is drowning in garbage PRs

Look at what major projects are doing:

  • curl — shut down its 7-year bug bounty program. AI-generated bug reports: only 5% were real. The rest was noise.
  • Ghost — outright banned AI-submitted code.
  • Tailscale — went further: closed all external PRs. AI or not, doesn't matter anymore.
  • GitHub itself — building a "PR kill switch" so maintainers can one-click disable external submissions.

Open source isn't being destroyed by hackers. It's being drowned to death by garbage AI-generated PRs from vibe coders.

Garbage in, garbage out

Here's the most ironic part of the whole chain:

You ask AI to write code → AI searches GitHub for references → It finds code written by other AIs → Garbage in, garbage out. What do you think comes out?

GitHub used to be a code repository. It's turning into a giant AI code dump.

The problem isn't AI. It's governance.

AI writing code isn't the problem. The problem is: nobody is auditing what AI does.

The core premise of vibe coding is "accept everything AI generates — don't review, don't modify." That works for a TODO app. For production code, it's planting time bombs.

But asking humans to line-by-line review AI code isn't realistic either. If 63% of devs say fixing is slower than writing, reviewing is even slower.

The real solution isn't "stop using AI" or "review everything manually." It's automated governance between AI code and production:

  • Every tool call is audited — every file change, every API call, cryptographically signed and recorded.
  • Dangerous actions are blocked — delete database? modify production? First pass through a 4-level decision tree. 97% auto-resolved, 3% escalated to human review.
  • Security policies evolve — every false positive, every bypass attempt feeds back into the governance engine.
  • Formal verification — not "we hope this is safe." Mathematically provable convergence toward safety.

That's MAREF. The open-source agent governance operating system. Brakes, seatbelts, dashcams, and a tireless security reviewer — for your AI agents.

Governance isn't a speed bump. It's what lets you go fast safely.

Un-governed AI code doesn't go faster. It just crashes faster.

The projects banning AI submissions — curl, Ghost, Tailscale — aren't anti-AI. They're saying the same thing: "We welcome AI assistance. We won't accept garbage without quality control."

MAREF is that quality control layer. Not between you and AI. Between garbage code and your production environment.

You can still vibe code. You can still have AI write your code. But before it deletes your database, someone will stop it.

You're getting slower? No. You're finally seeing what you're actually shipping.


📊 Sources: GitHub Octoverse Report, Carnegie Mellon fake star study, Socket security report, CodeRabbit 470 PR analysis, Meter RCT trial, curl/BT/Ghost/Tailscale official announcements. MAREF is an open-source agent governance operating system. Get started in 5 minutes.