Hacker News

Article URL: https://www.reuters.com/business/media-telecom/us-supreme-court-wont-hear-metas-challenge-vermont-social-media-addiction-2026-05-26/

Comments URL: https://news.ycombinator.com/item?id=48283391

Points: 2

# Comments: 2

Article URL: https://rev.omniiii.com/app/6499478532?country=us&agg=false&ratings=%5B%5D&sentiments=%5B%5D&themes=%5B%5D&q=&sort=newest&bucket=week&mode=pct

Comments URL: https://news.ycombinator.com/item?id=48283376

Points: 1

# Comments: 0

Article URL: https://github.com/lwouis/alt-tab-macos/discussions/5533

Comments URL: https://news.ycombinator.com/item?id=48283349

Points: 2

# Comments: 0

We built FlowLink because AI agents (Claude Code, Cursor, Copilot) keep executing destructive commands on production servers with no guardrails.

The recent "AI agent deleted production database" post (860 pts here) is exactly what this prevents.

FlowLink is an MCP proxy that sits between your AI agent and your tools. No code changes required. Point your agent config to FlowLink and it starts intercepting destructive commands.

What it does:

Shield Engine intercepts rm -rf, DROP TABLE, git push --force, chmod 777 and 100+ destructive patterns BEFORE execution.

Policy Engine: per-agent, per-tool rules (e.g. "Claude can read but not delete").

Zero-Trust Secrets: agents get scoped, time-limited tokens, never raw credentials.

Telegram approval queue for human-in-the-loop on high-risk operations.

Full audit trail of every agent action.

Setup takes 2 minutes. Works with Claude Code, Cursor, Copilot, any MCP-compatible agent.

Tech: Rust backend, MCP protocol native, E2EE, self-hosted.

Website: https://flowlink.flow-masters.ru

What guardrails are you currently using for AI agents in production? What's missing?

Comments URL: https://news.ycombinator.com/item?id=48283348

Points: 1

# Comments: 0

Article URL: https://knowwhereconsulting.co.uk/blog/shakespeares-world/

Comments URL: https://news.ycombinator.com/item?id=48283312

Points: 2

# Comments: 0

Article URL: https://www.theverge.com/news/937221/nvidia-geforce-control-panel-app-retirement

Comments URL: https://news.ycombinator.com/item?id=48282670

Points: 1

# Comments: 0

Article URL: https://sosuke.com/you-can-start-building-llm-skills-before-you-know-the-whole-shape/

Comments URL: https://news.ycombinator.com/item?id=48282663

Points: 1

# Comments: 1

Article URL: https://www.normaltech.ai/p/do-ai-risks-require-extraordinary

Comments URL: https://news.ycombinator.com/item?id=48282659

Points: 1

# Comments: 0

Article URL: https://github.com/Dylanchess0320/LuckyD-Code

Comments URL: https://news.ycombinator.com/item?id=48282639

Points: 1

# Comments: 0

Article URL: https://languagetool.org/webextension/premium-announcement

Comments URL: https://news.ycombinator.com/item?id=48282635

Points: 1

# Comments: 0

Article URL: https://www.nklswbr.com/blog/new-is-always-better

Comments URL: https://news.ycombinator.com/item?id=48282626

Points: 1

# Comments: 1

Article URL: https://github.com/TangibleResearch/Halgorithem-Playground

Comments URL: https://news.ycombinator.com/item?id=48282616

Points: 1

# Comments: 0

Article URL: https://davivek.github.io/posts/dont-migrate-to-find-out/

Comments URL: https://news.ycombinator.com/item?id=48282615

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=XKxRjFnOxPc

Comments URL: https://news.ycombinator.com/item?id=48282607

Points: 1

# Comments: 0

Article URL: https://element.io/blog/sweden-goes-live-with-matrix-based-federation/

Comments URL: https://news.ycombinator.com/item?id=48282574

Points: 1

# Comments: 0

Speakrs implements the full pyannote community-1 style diarization pipeline in Rust: segmentation, powerset decode, overlap-add aggregation, binarization, embedding, PLDA, and VBx clustering.

There is no Python runtime in the library path. Inference runs on ONNX Runtime or native CoreML, and the rest of the pipeline stays in Rust.

It is 20x-30x faster on macOS, but only 2-3x faster on linux/cuda (depending on CPU).

Few reasons its faster:

1. Speakrs is using coreml versions of the models. I exported the models specifically to run on coreml. PyAnnote just runs the same the same PyTorch versions through MPS (Metal) on macOS.

2. PyAnnote is not a single model, its a few different models put together in a pipeline, the readme has some info on the full pipeline.

3. Speakrs optimizes the pipeline so different parts can run on CPU, Neural Engine and GPU. Speakrs has a batch mode, where you can run on multiple files at once, doing this also lets you keep CPU/GPU/ANE all fully utilized.

This is why on linux/cuda its not that much faster, PyAnnotate is already optimized to run on cuda, the speed improvements we get on cuda is by running some stuff on cpu while the other stuff runs on the GPU. The speedup on linux will depend on how powerful the CPU is.

There is also a fast mode, that sacrifices some speed for accuracy, that can be up to 50x faster, and for some types of audio doesn't sacrifice that much accuracy. The benchmarks have more info on this.

Comments URL: https://news.ycombinator.com/item?id=48282551

Points: 2

# Comments: 0

Article URL: https://www.theglobeandmail.com/business/article-artificial-intelligence-business-creation-us-canada/

Comments URL: https://news.ycombinator.com/item?id=48282549

Points: 2

# Comments: 0

Article URL: https://undark.org/2026/05/21/opinion-hantavirus-medical-mistrust/

Comments URL: https://news.ycombinator.com/item?id=48282546

Points: 1

# Comments: 0

Article URL: https://freakonomics.com/podcast/in-a-driverless-world-who-loses-and-who-wins/

Comments URL: https://news.ycombinator.com/item?id=48282543

Points: 1

# Comments: 0

Article URL: https://contraptions.venkateshrao.com/p/commodity-intelligence

Comments URL: https://news.ycombinator.com/item?id=48282532

Points: 1

# Comments: 1