ATechGuy

Interesting. Care to list them here so that we all can learn.

Have you looked into open-source alternatives? I'm assuming that it's a pressing problem for you, and you have already explored alternatives.

Are you sure that CF can stop AI bots?

Just saw this https://x.com/ycombinator/status/1960779353589211577

They say "... can scrape any website—not even Cloudflare can detect it."

Mind sharing what kind of mitigations you put in place and how well they worked?

Great! AI crawlers would love this.

Thanks!

Thanks for sharing the links! You're right, the legality around scraping is nuanced :( Use cases like license compliance might fall on the defensible side.

Right now the cost is a blocker (I'm bootstrapped, no funding yet). On top of that, sometimes official APIs either don't exist or don't expose the full dataset needed for license checks, which makes things tricky.

Looks like he has cracked the hiring playbook. I wouldn't be surprised if Zuck came forward and said they also hired SP for their ASI team.

Honeypots may be skipped by bots. My simple solution to this is to create an input/text field in the account creation form, add a placeholder text "I am human" and ask the user to enter what they see (i.e., placeholder text). I've never received any fake/spam account since I incorporated this technique.

> At the moment, it seems Basic mode is so basic that it allows everything to pass as human. That’ll likely change as they gather more telemetry to better identify what a bot signal looks like.

So they are basically collecting telemetry in the name of "free basic anti-bot" solution.

Some of the best decisions I made ever: 1. Deleted FB in 2012 :) 2. Didn't create insta or whatsapp account 3. Never applied to meta jobs

Please don't deploy this on the internet, it may block real users and lock them out.

> There is nothing anyone can do to prevent bots.

Are you sure about this?

If it does not cost them everything, they will not stop.

Unfortunately, everything you mentioned above is a possibility.

Has someone noticed a similar thing with ChatGPT and private Github repos? ChatGPT has recommended private repo links to me many times. Because they are not public, I get repo not found error. But ChatGPT can generate private code with no issues.

Congrats on launching! Booting VMs in milliseconds is certainly important, but it can also be achieved with CloudHypervisor/Firecracker. Where Containers beat VMs is runtime perf. The overhead in case of VMs stems from emulation of IO devices. I believe the overhead will become noticeable for AI agentic use cases. Any plans to address perf issues?

Are you looking for an open-source sandboxing solution? Self hosting is available for E2B. You still have to subscribe to a SaaS for ephemeral cloud compute though.

This. Big tech providers already offer confidential inference today.

Thanks! This is a very detailed explanation of why existing sandboxing techniques will not work as expected for dependencies (wrt to functionality or performance).

I believe enterprises that care about privacy are using private AI from big tech (say Github copilot), others may not care so much about it.

That was my question too.

Have you tried using private inference that uses GPU confidential computing from Nvidia?

Why not use confidential computing based offerings like Azure's private inference for privacy concerns?

Can you execute Python code under WASM? If so, how stable is that?

Looks like a useful tool. Congrats on shipping! Many packages are installed automatically in environments like CI/CD pipelines or Dockerfiles, where interactive review and consent aren't possible. How do you plan to handle such scenarios?

Sorry if I'm missing something, but why can't Anubis not detect/block such AI agents? Is it because they use headful browsers?

Can browser-use be blocked using Anubis or other anti-bot measures?