DelightOne

Many people don't wanna depend on US products. If US switches them off, you are gone. Not even talking about privacy of US companies. Nor talking about guardrails that let you not do things you need to do.

1) What kind of guardrails are you using? Do you do models like prompt injection checkers? Or what guardrails are useful that do not know the environment the agent lives in?

2) I heard Bifrost also works for this. Did you try that as well, pros/cons?

Are there good open source setups that generate this training data automatically?

Or is that the secret sauce no one wants to share, the edge people see themselves having.

If it costs you as good as nothing, you might as well do it.

Previously, time was the most precious resource. Now its tokens, and more cheaply at at.

Its easier for an LLM to find vulnerabilities in a project with less usage, and those vulnerabilities will stay open longer, making them much cheaper to attack to keep the door open.

It also means security is not held as high and vulnerabilities not as much found. A simple 0-day may survive for years. Not much effort needed to have permanent access.

Did anyone do prompt injection detection?

Prefix caching automatically detects priorly read tokens -> more fained grained checkpoints then. With prefix caching you don't have to ask for a checkpoint, its automatic.

One way or the other, a prefix is saved. Only the additional info how to continue is new.

With prefix caching, you get checkpoints for free. Do you mean that?

Are you able to just re-run the data for the timeframe, or is "fixing it" a more manual process? And what observability are you working with into the process?

That's what install outside of the App Store is for. On your own risk-

I don't want random apps to paste potentially dangerous things into other apps. Its understandable.

Imagine a banking app, and for example an IBAN field.

So if they played a short annoncement beforehand so people know its an Original, it would be fine? Originals get advertised heavily, next-movie, so I assume putting it in the same playlist is fine.

Amazon Originals, Netflix Originals. Disney Originals. Paramount Originals. I'm just wondering what is different between series and music, that for music its very bad morally to create your own and to put your own in the front row. While for other streaming its accepted.

Isn't that how every streaming service does it?

In what environment do you run such tests? Do you have a script for it, or do you have a UI that manages the process?

> This time, I tried to learn from that: facts are stored as instants, reasoning happens in local days of the jurisdiction that cares.

I think that's how the JavaScript Temporal proposal works. Convert your instant to the timezone, make the comparisons/calculations, hope you didn't jump an hour due to summertime, convert back.

They don't wanna be like Facebooks' .1%.Thy know your user.

The threat already silences the opposition. You don't have to use it to silence people.

I bet they tried an llm scam detector. It was too good so they didn't put it into production.

Its death by a thousand cuts.

Making people able to sue for anyone feeling bad about not having gotten the job is a path you should not take. We have something similar in Germany and its horrible for companies. Leeches bleeding you dry.

How does an e2e test for less capable LLMs look like, you call each LLM one by one? Aren't these tests flaky by the nature of LLMs, how do you deal with that?

Congratulations!

I've got a small question. How do you deal with people asking for open sourcing your product/code, claiming they don't want to use a product they don't control?

> And be focused. Understand everything in every article, do not accept not understanding everything.

I read my first programming books that way. Always ask "Why is that word here" and "Why is that ! there". Made me see details.

Every time I hear it contains AI it sounds like the features' outcome will be uncertain. Especially with more experience. If your feature were good you wouldn't have to mention AI to show it is awesome.

More than not AI is used as an excuse for the feature to be bad.

Problem is most innovation is in using existing models in new ways. You can't expect these most people to train their own models.

Regulating it the way you say just means "zero innovation!".

The UX is not good enough yet. We would have to 1) show that it is reserved to people going on the link and 2) offer good/enough ways to be notified once the link becomes online and 3) need to know the likelihood of the link to actually work in the future based on prior commitment.

AND I probably forgot a couple of issues.

Or sync it with Dropbox. Just drop newly-rendered tables into the images folder and bam! you can directly reference them in your latex.

No wonder Sony goes after Windows people. Half of them soon need a new device.