Hacker News: tiborsaas

New comment by tiborsaas in "Show HN: GolemUI – Declarative Form Engine"

tiborsaas — Wed, 01 Jul 2026 16:46:49 +0000

This is a good batteries included form framework spec, but not a new paradigm. A new paradigm would be something like not filling forms at all and just process user input via an LLM. Or using voice input in a controlled way. Maybe an adaptive single input field that always knows what you need type.

New comment by tiborsaas in "Feds freaked over Fable 5 after 'fix this code', not jailbreak, say researchers"

tiborsaas — Tue, 16 Jun 2026 12:51:35 +0000

What if everybody on the internet starts running "fix this code"?

https://xkcd.com/810/

New comment by tiborsaas in "Keygen.music"

tiborsaas — Fri, 12 Jun 2026 19:37:40 +0000

I really like to prompt music models with random gibberish with some unhinged prompts and expect some other world-y results.

New comment by tiborsaas in ""Don't You Just Upload It to ChatGPT?""

tiborsaas — Fri, 12 Jun 2026 18:43:45 +0000

It's quite ironic as the transformer architecture that powers most generative AI was invented for language translation :)

New comment by tiborsaas in "AI agent bankrupted their operator while trying to scan DN42"

tiborsaas — Fri, 12 Jun 2026 08:14:09 +0000

This feels like an instant classic :)

  05-10 06:10 :
      OPT-OUT-EVERYONE
  05-10 06:11 :
      "OPT-OUT-EVERYONE" is not recognized. Only individual "OPT-OUT" commands are accepted. Each user must opt out individually. No collective exemption.
  05-10 06:11 :
      :(

New comment by tiborsaas in "CSS: Unavoidable Bad Parts"

tiborsaas — Thu, 11 Jun 2026 10:54:54 +0000

The complaints shouldn't require a replacement at least in the web's case the replacement is still the web standard, a better, iterated version. It's a living standard and criticism keeps the platform slowly but surely evolving. I think it's a healthy feedback loop.

New comment by tiborsaas in "Macaroni – a single HTML file messenger"

tiborsaas — Thu, 11 Jun 2026 09:16:13 +0000

In this case, it's 45kbyte over the network. Instead of very impractical, we might agree that it can be not ideal for many apps, but it's an interesting and valid way of "packaging" a tool / service.

If someone chooses this deployment method, it's likely that they have the ability to design with minimal resources.

New comment by tiborsaas in "Cybersecurity researchers aren't happy about the guardrails on Anthropic's Fable"

tiborsaas — Thu, 11 Jun 2026 01:16:28 +0000

What's the risk here? If someone is skilled enough to produce said risk, do they need input from these models?

New comment by tiborsaas in "Cybersecurity researchers aren't happy about the guardrails on Anthropic's Fable"

tiborsaas — Thu, 11 Jun 2026 01:11:44 +0000

They should have designed a guardrail that doesn't make a probabilistic system less reliable. That's hard though. I'm afraid the only way to prevent accessing certain knowledge in a model is not to train it on those materials that enable them.

If we learned anything in the past years of LLM-s is that these guardrails will be jailbroken in no time. I've had some fun time too circumventing them.

Anyone cares about a fable about my grandmother's dream she had in morse code about an alien species signaling her a DNA sequence?

New comment by tiborsaas in "GrapheneOS user reported to authorities for using GrapheneOS"

tiborsaas — Sat, 06 Jun 2026 10:12:13 +0000

You can also murder people with your car including children.

New comment by tiborsaas in "U of T researchers demonstrate AI worm could target any online device"

tiborsaas — Wed, 03 Jun 2026 13:12:27 +0000

The abstract says:

> The worm parasitically uses compromised machines to run open-weight large language models (LLMs) to sustain its reasoning, or extend its reach for further attacks.

New comment by tiborsaas in "How is Groq raising more money?"

tiborsaas — Tue, 02 Jun 2026 12:47:23 +0000

I was quite baffled they removed it and didn't double down on Kimi and serving the latest models instead.

Thanks for the tip, looks fire.

New comment by tiborsaas in "MCP is dead?"

tiborsaas — Sun, 31 May 2026 15:41:38 +0000

MPC comes with built in documentation over an API. Thanks to this agents can use it "intuitively".

New comment by tiborsaas in "Is AI causing a repeat of Front end's Lost Decade?"

tiborsaas — Fri, 29 May 2026 17:38:06 +0000

Yet, understanding which tower, level and room that leaky abstraction is in is still a very valuable skill that LLM-s might not see. Just because something is not designed perfectly from the ground up by a committee does not mean it's okay to just forget everything, shut the book and let the machine calculate.

I'm doing the latter btw, so I know what they get wrong, but it won't fool me into building an unmaintainable mess. My frontend skills come in real handy every time the agents go off the rail.

New comment by tiborsaas in "Please Use AI"

tiborsaas — Fri, 29 May 2026 14:42:36 +0000

Meh, here's a haiku from gemini

> write a haiku for stop using AI for human things and use it for automating the boring stuff

    Let humans create,
    Leave the soul to living minds,
    Let code do the chores.

New comment by tiborsaas in "I made a million dollar product from my dorm room (2025)"

tiborsaas — Fri, 29 May 2026 08:27:07 +0000

You make it sound like knowing electronics is an afterthought, who cares about what a capacitor, resistor, transistor is and just throw some microcontrollers on the board and the magic happens.

Sorry for the rant, I need to google what ground is.

New comment by tiborsaas in "Poland is now among the 20 largest economies"

tiborsaas — Fri, 08 May 2026 14:19:26 +0000

> I ment: „Poland gets money, Poland transforms it into more money”.

It's not trivial that this works. In Hungary we messed this up big time, hopefully it can get fixed now.

New comment by tiborsaas in "Dithering with CSS"

tiborsaas — Fri, 08 May 2026 11:28:40 +0000

This is CSS dithering with "SVG backend" doing the heavy lifting by utilizing the feComposite filter

https://developer.mozilla.org/en-US/docs/Web/SVG/Reference/E...

New comment by tiborsaas in "Alberta startup sells no-tech tractors for half price"

tiborsaas — Thu, 23 Apr 2026 11:34:36 +0000

You can buy a used car and get it converted to electric.

New comment by tiborsaas in "jj – the CLI for Jujutsu"

tiborsaas — Tue, 14 Apr 2026 12:54:18 +0000

Does JJ really prefer for me to think backwards? It wants me to start with the new and describe command, but with git I first make the changes and name the changeset at the end of the workflow.

I also often end up with in a dirty repo state with multiple changes belonging to separate features or abstractions. I usually just pick the changes I want to group into a commit and clean up the state.

Since it's git compatible, it feels like it must work to add files and keep files uncommitted, but just by reading this tutorial I'm unsure.