The word "accidentally" is slippery, our understanding of how accidents can happen with software systems is not applicable to LLMs.

> No internet open browsing or open chat features. > AI toys shouldn’t need to go online or talk to strangers to work. Offline AI keeps playtime private and focused on creativity.

> No recording or long-term data storage. > If it’s recording, it should be clear and temporary. Kids deserve creative freedom without hidden mics or mystery data trails.

> No eavesdropping or “always-on” listening > Devices designed for kids should never listen all the time. AI should wake up only when it’s invited to.

> Clear parental visibility and control. > Parents should easily see what the toy does, no confusing settings, no buried permissions.

> Built-in content filters and guardrails. > AI should automatically block or reword inappropriate prompts and make sure results stay age-appropriate and kind."

Obviously the thing users here know, and "kid-safe" product after product has proven, is that safety filters for LLMs are generally fake. Perhaps they can exist some day, but a breakthrough like that isn't gonna come from an application-layer startup like this. Trillion dollar companies have been trying and failing for years.

All the other guardrails are fine but basically pointless if your model has any social media data in its dataset.

The kind of obvious propaganda like "it's a crime to have the same haircut as Kim Jong-Un" (or to not have it, depending on the source). No one is saying life there is great, but there is a track record of fantastically untrue stories.

What if I told you that's not the key question, and the "more data" approach has obviously and publicly hit a wall that requires causal reasoning to move past?

How does your library work in this regard? If my node server is down, will my scheduled tasks still execute? I notice you have a .start() method, what does that do? Is it polling periodically?

There is a pretty big underlap between these problems, and the ones solved by SpacetimeDB. So much so that if you built a system using both SpacetimeDB and another, high-level tool to manage chunking, transitioning, server provisioning etc. it would obviously be the latter system that was more indispensable to scaling your MMO.

I kept on reading expecting to see motion prediction, multiversion, or similar given the name and focus on games, but no. This is a totally normal database, designed for low latency and with support for WASM stored procedures. You can host your own server or they will rent you one. Don't get me wrong, this looks very nice. It looks like a solid building block for persistent worlds in multiplayer games. You'll just have to do your lag masking netcode yourself.

https://news.ycombinator.com/item?id=43593913

>create_table

Commands are not declarative. You want a single file with the state of your schema in it? We call that a schema file. It is strictly a negative to declare your schema in SQL, an imperative language. There is no way to guarantee this create_table will run safely, because it's not meant to ever be run.

Any plans on bundling a local LLM / supporting local LLMs?

If only there was some middleground between "no remote code execution" and "breaking the entire paradigm of adblockers".

The statement that they could be "rewritten" to anywhere near the same effectiveness is straight up disinformation. There are years and years of coverage of this issue that contradict you. The debate has nothing to do with the level of effort for rewrites. This is a hostile decision by a company with a directly adversarial relationship to adblockers.

I gotta tell you, a typo in the name of your product on your homepage makes me question if you're ready for prime time.

Solutions like lavamoat have been proposed and implemented in places, but they will not scale to the entire industry. There's a bigger problem in how innovation speed / developer convenience is inversely related to security.

In many cases it's literally cheaper to just wait for an attack and make victims whole. And even if not, it's hard to explain to your investors why it takes you half a year to build a product that took your competitors a month, so you just pretend the threat doesn't exist.