Yes, the OpenAI disclosure about the same incident is much better https://openai.com/index/mixpanel-incident/

Much better ?

What to know about a recent Mixpanel security incident Transparency is important to us...

They're so much transparent that they leaked PII to Mixpanel...

Same for CoinTracker; more detailed than the original -- https://news.ycombinator.com/item?id=46065208

HN discussion of OpenAI’s notice about this Mixpanel situation:

https://news.ycombinator.com/item?id=46065585

> Has Mixpanel been removed from OpenAI products?

    Yes.

I'd be more interested in understanding why OAI would think exporting PII to a 3rd party platform was acceptable. As for whether they follow the same standard with other providers, all bets are now off

At Kilo (VScode plugin, BYOK so works with Claude Pro) we just rolled out autocomplete to 10% of our user base on Friday and we're happy with what we're seeing. The autocomplete is based a Mistal model (Codestral) that is fast and accurate https://mistral.ai/news/codestral We copied a lot of code from Continue.dev (best open source autocomplete in our view). We plan to roll it out to 100% of our users no later than Tuesday. The autocomplete is consumption priced (no markup) but the cost per month should be low single digit dollars.

Kilo’s been on my list of tools to try anyway. Good to hear you’ve got a strategy for useful autocomplete. Any idea how it compares with Supermaven both in terms of speed and quality? Supermaven really had found a sweet spot for both. I found LLM autocomplete was generally more annoying than it was worth prior to Supermaven.

Sorry but I'm don't know how Kilo autocomplete compares with Supermaven. Maybe someone else does?

BTW For completeness, while we’re still in rollout for existing users autocomplete is default on for all new users already.

It doesn't look near instant like Supermaven because the default delay before suggestions is 3 s.

https://kilo.ai/docs/basic-usage/autocomplete

The cost improvements are great. If you miss the automation that AWS does for database servers consider using something like https://www.ubicloud.com/ that is great for PostgreSQL servers. On bare metal these typically also support 5x the number of IOPS without paying through the nose.

So the core idea is to use an LLM to draft reasoning as a structured, JSON domain-specific language (DSL), then deterministically translate that into first-order logic and verify it with a theorem prover (Z3).

Interesting that the final answer is provably entailed (or you get a counterexample), instead of being merely persuasive chain-of-thought.

I should have hit Fort Bragg by now but I'm not seeing that on https://www.noyocenter.org/live-web-cam

The earthquake was real but it is probably really hard to predict if that leads to a tsunami.

I was watching the crescent city webcam and watched the fishing boats rushing out of the harbor after the quake. The webcam is now down, so I hope they made it.

Crescent city has a long history of being devastated by norcal quakes due the the bathymetry and costline profile.

Is that webcam guaranteed to be a live/up to date feed? regardless I'd imagine there to be some confirmation by now.

There's a timestamp in the top right, so yes. Looks like no tsunami

Direct link to their live cam on YT:

https://www.youtube.com/live/ESsx4MdloQw?si=PQSwV06GsYfwOZPj

Cool to see that Google Docs gives me this warning too https://imgur.com/a/B6qSboV

That's it? "Tsunami Warning", no context? no advice on what to do to avoid panic?

I would suggest that this is based on location of the IP address. I'm not getting it on the other side of the world.

Presumably a person receiving such an alert would have a heightened sense of self preservation and would investigate further..

Congrats on launching Oege!

This is the best article on the problems of airborne wind energy. It is 10 years old but I think Michael Barnard did a great job explaining all the variants and why it is hard https://czmphorfqgktmzxu.quora.com/Airborne-wind-energy-a-co...

You were not kidding. That article really covered a lot of the obvious questions I had (and was the single best thing I've ever read on Quora).

HugOps to the Drew, the rest of the Sourcehut crew, and anyone else working on this. Compliments on the clear blog post.

'Inmiced' lol! Thanks for keeping the titles accurate, so important especially with cancer research results. Happy holidays and I really enjoyed reading the quote on your profile https://news.ycombinator.com/user?id=dang

It's the least of your accomplishments but thanks for being such a great HN contributor over the years!