Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Grok Bot

x.ai

Post reply on HN
Print view View original

321–330 of 362 posts

Re: Grok Bot

#321

Post by gexla »

This is super confusing. It's not supposed to have a Linux version and it's supposed to be only for Supergrok Heavy (during beta according to the docs) and yet the landing page gave me an option to download it for Linux and to setup a trial account that I could use it with via Cursor. But, it doesn't seem to get through the gates to connect to whatever it needs to connect to.

Otherwise, I think this would still go into my "I don't get it" folder. It seems like the flavor of the day is "agents can talk to each other" and Grok Bot is selling that point along with Claude Code. But I think Pi has had this for a while? Or it was easy to wire it in.

Re: Grok Bot

#322

Post by diseasedyak »

therealdrag0 wrote: post #180
whimsicalism wrote: post #62

How do they stop providers (like Amazon, etc.) from detecting and blocking these agents if they are running on cloud? I know that openai wasn't able to avoid this which is why they moved to 'computer use' on your local machine. I almost wonder if this is a place where SpaceX, as an internet provider through starlink, has a unique advantage because websites are unable to block their networking as it could be residenti…

I use Hermes locally and it’s constantly hitting bot blocks. Just trying to shop clothes for me it gets blocked.

This exactly! It sounds like a good thing in theory, but in practice everything is blocked that I've ran into, using Hermes to try to automate things like this. It's not just bot blocks either, but things like Captcha and such.

Re: Grok Bot

#323

Post by diseasedyak »

I don't understand how they get around bot blockers and captcha. I use Hermes locally and you can't get around any of that, even using your own credentials.

Re: Grok Bot

#326

Post by keeda »

ofjcihen wrote: post #255
shaism wrote: post #49

Earlier quoted context omitted.

According to Boris Cherny from Anthropic [1], the threat of prompt injection has been largely solved. [1]: https://x.com/bcherny/status/2086520950259118464

As someone who just got out of a meeting demonstrating how Copilot running Luna can be breadcrumbed by a single line of text in innocuous package into downloading and installing malware I think Boris Cherny might be wrong.

Boris is talking about Anthropic models only of course. In the chart in his tweet Luna has like a 43% chance of being compromised with sufficient attempts. But even the Anthropic models have as high as a 5% chance, which is still far too much IMO.

Re: Grok Bot

#327

Post by mrguyorama »

redox99 wrote: post #139
dylanhouli wrote: post #101

I feel like we'd be better off if we just stopped at chatbots...why are we so eager to make the internet even more botted

The internet sucks. Yesterday, I had my agent search for openings for The Odyssey that fit my requirements and then book them. It was way better than manually looking at seat maps for 20 different showtimes and going through 10 steps just to buy the tickets.

This is functionality currently built into Google search. I google "The Odyssey Showtimes" and it shows me when it's playing at all local theaters.

Then I can just pick which time works for me. I have never in my life had a theater be sold out when it's not an opening night or weekend for extremely big films.

Is this something you actually tried to approach before LLMs?

Re: Grok Bot

#328

Post by redox99 »

mrguyorama wrote: post #327
redox99 wrote: post #139

Earlier quoted context omitted.

The internet sucks. Yesterday, I had my agent search for openings for The Odyssey that fit my requirements and then book them. It was way better than manually looking at seat maps for 20 different showtimes and going through 10 steps just to buy the tickets.

This is functionality currently built into Google search. I google "The Odyssey Showtimes" and it shows me when it's playing at all local theaters. Then I can just pick which time works for me. I have never in my life had a theater be sold out when it's not an opening night or weekend for extremely big films. Is this something you actually tried to approach before LLMs?

They weren't sold out, they had bad seats or didn't have 4 contiguous ones.

Re: Grok Bot

#329

Post by slg »

brabel wrote: post #285
inigyou wrote: post #272

Earlier quoted context omitted.

Besides technological progress which has nothing to do with this effect and would have happened anyway, what's the value that we didn't have 30 years ago?

It’s so incredible to me that now we have a chat interface we can ask about anything in any language and get really great answers, something literally considered science fiction a few years ago, and people still act like that is no big deal at all. No value in something like that! Don’t tell me it’s inaccurate, I strongly believe it’s way more accurate than if you could ask an expert in each topic , which of course y…

>It’s so incredible to me that now we have a chat interface we can ask about anything in any language and get really great answers, something literally considered science fiction a few years ago, and people still act like that is no big deal at all.

Because the importance of this is all about perspective. It wasn't like these systems created this information out of thin air. They were trained on something. That means the answers they are giving you have been available for decades. You just needed the know-how to find that information and synthesize the answers yourself. To many of us, it's like going from the old physical card catalogs to a modern digital system that would have seemed like sci-fi to a prior generation too. It's definitely more efficient and easier to use, but people acting like it's revolutionary seem to be suggesting that the old system didn't exist or wasn't usable with a little effort.

Re: Grok Bot

#330

Post by ofjcihen »

keeda wrote: post #326
ofjcihen wrote: post #255

Earlier quoted context omitted.

As someone who just got out of a meeting demonstrating how Copilot running Luna can be breadcrumbed by a single line of text in innocuous package into downloading and installing malware I think Boris Cherny might be wrong.

Boris is talking about Anthropic models only of course. In the chart in his tweet Luna has like a 43% chance of being compromised with sufficient attempts. But even the Anthropic models have as high as a 5% chance, which is still far too much IMO.

That’s fair and to their credit (and somewhat the user’s detriment) I think the user focused guardrails they’ve imposed have unintentionally helped to protect against the kind of attacks I have success with.

But you’re right…5 percent is better than what it was but at scale that’s still quite a bit.

Post reply on HN