Saw this today https://news.ycombinator.com/item?id=42575951 and thought that there might be more such cautionary tales. Please share your LLM horror stories for all of us to learn.
Stories by ATechGuy
Seeing a lot of people running coding agents (Claude Code, etc.) in custom sandboxes Docker/VMs, firejail/bubblewrap, scripts that gate file or network access.<p>Curious to know what's missing that makes people DIY this? And what would a "good enough" standard look like?
As software providers, we use several FOSS libraries. These libraries require periodical security bug fixing, which requires shipping a new release. We are building an LLM agent that can block these security bugs at runtime (like a firewall so to speak) and prevent attackers from exploiting security...
Sometime we need to execute LLM generated code from within our app. For that we are using VMs today, but it's slow. How are others handling this?
What has been your experience with dev environments? We understand the pros, but are there pitfalls we should know before making the transition?
ReCAPTCHA is used on millions of websites to verify human users. Given the massive scale and visibility of these widgets, it looks like a potential goldmine for impressions. Yet, Google has never added ads to it.<p>Considering their focus on ad revenue, this seems like a missed opportunity, or maybe...
Curious how people's views on AI/LLMs have evolved over the last couple of years.
I've noticed several AI bots crawling my website. Is there a way to figure out which specific user queries caused the AI to access my website?
Do you use Docker containers or VMs to securely execute code generated by AI agent? What are pros/cons of your setup? Appreciate your insights.
Have you made (or would like to make) any customizations to Firecracker or Gvisor for your use case? If so, please share.
In the last couple of months, several new solutions for sandboxing AI agents have launched (microVMs, WASM runtimes, browser isolation, hardened tool containers, etc.). Curious to hear from people using them in production. Are they working as advertised, or are there still major tradeoffs around sec...
The founder of Groq recently made the case that increasing compute is the key to advancing AI capabilities. Curious to hear your thoughts.
I'm working on a service for FOSS developers to help enforce code license compliance and make projects more sustainable.<p>The challenge: many websites' Terms of Service explicitly prohibit scraping, crawling, or automation. At the same time, the information needed (repos, dependencies, me...
We're building an AI agent for a niche use case. What tools exist today for agents to debug, test, rollback, etc.
Those who are using Firecracker or Gvisor, what are your use cases? Have you made (or would like to make) any customizations?
Hi HN,<p>We are launching serverless Python workers for developers who are building Python web/AI apps: <a href="https://pyworkers.dev" rel="nofollow">https://pyworkers.dev</a><p>Our Python workers similar to Cloudflare workers, except for Python and are long running, which ...
Lately we've been noticing a non-trivial amount of traffic in our logs that doesn't look like typical bots.<p>Not the usual noisy crawlers or obvious scrapers. The behavior is different with fewer hits, more selective page access.<p>Some of the user agents suggest AI crawlers, but some do ...
[Posting again in hopes of gaining further perspectives and insights]<p>Do you still browse the web the same way you did (say 5 years ago) or you use AI to browse, summarize, etc. on websites? Is it about speed/efficiency, instant answers (gratification), or something else?<p>I've noticed ...
Have been thinking about this a lot lately: assuming AI advances to a level where it can autonomously operate as <HIRED-FOR-ROLE> with less than 1% errors, would you replace it with AI? Why or why not?
YC has recently published a list of startup ideas they are interested in funding. You can check it out here: https://www.ycombinator.com/rfs. What do you think about it, and would you consider developing one of these ideas?
Are you open to using AI agents for your everyday repetitive tasks like ordering coffee? If so, what use cases you'd want the agents to address?
Do you still browse the web the same way you did (say 5 years ago) or you use AI to browse, summarize, etc. on websites? Is it about speed/efficiency, instant answers (gratification), or something else?<p>I've noticed I'm less willing to read long articles now that I can get instant s...