I definitely would’ve considered Tesla as my first EV but as of now they’re dead to me. If he was completely gone then that actually becomes a selling point for me.

“Merit-based” hiring.

I imagine an ad blocker could prevent this data going out, unless the hosts were generic and the game/app simply won’t work without allowing those connections. I’ve never seen an app be [obviously] broken from my ad blocker but I am interested in running a similar experiment to see just how much data is going out.

So rather than build off their successes we’re just gonna put up a blinder and hope they don’t completely leave us in the dust?

I like named volumes, externally created, because they are less likely to be cleaned up without explicit deletion. There’s also a few occasions I need to jump into a volume to edit files but the regular container doesn’t have the tools I need so it’s easier to mount by name rather than hash value.

Also check out https://buynothingproject.org/

To be clear, this page is a lie? https://proton.me/security/end-to-end-encryption

So even if I have the recipients public key the message actually goes to Proton servers in plaintext before it is encrypted?

The contents can be read, sure, but unless ChatGPT is doing a lot of hallucinating at least a few providers support e2e encryption and don’t manage the private keys.

Edit: To avoid reading the whole thread, providers may support E2EE but can’t guarantee it in all cases. A guarantee requires the clients on each end to manage the encryption and decryption so no plaintext enters the network.

Can’t remember what big platform provider, maybe Google, was giving out personal info to the government just because they were asked (i.e. no warrant given).

My Black & Decker Matrix drill. I have 2 other attachments for it and I could see myself getting as many as 8 of them. I’m sure they’re sub-par compared to their standalone counterparts but I’m not a hobbyist so I’d much rather have versatility and not have to make room for a chest full of tools.

Same with my cast iron pans.

It’s the perfect amount of instant coffee!

Yup, we can’t continue forever with educated but low-level workers getting 3% raises when the CEO gets a 30% raise.

“Waaaaah” you say?

For gay men, where hats and wear clothes that kind of clash. Drab colors too. My fave is grey shirts and grey pants.

It’s up to the user to understand it’s a fantasy and not reality.

I believe even non-AI media could be held liable if it encouraged suicide. It doesn’t seem like much of a leap to say, “This is for entertainment purposes only,” and follow with a long series of insults and calls to commit suicide. If two characters are taking to each other and encourages self-harm then that’s different. The encouragement is directed at another fictional character, not the viewer.

Many video games let you do violent things to innocent npcs.

NPCs, exactly. Do bad things to this collection of pixels, not people in general. The immersion factor would also play in favor of the developer. In a game like Postal you kill innocent people but you’re given a setting and a persona. “Here’s your sandbox. Go nuts!” The chat system in question is meant to mimic real chatting with real people. It wasn’t sending messages within a GoT MMO or whatnot.

Llms are quickly going to be included in video games and I would rather not have safeguards (censorship) because a very small percentage of people with clear mental issues can’t deal with them.

There are lots of ways to include AI in games without it generating voice or text. Even so that’s going to be much more than a chat system. If Character AI had their act together I bet they’d offer the same service as voice chat even. This service was making the real world the sandbox!

The failure is reasonable scenarios where the fantasy needs to end. AFAIK the only other way this could’ve ended, without harm, would be if the kid just decided to stop chatting (highly unlikely) or if someone looked over his shoulder at what was being typed (almost as unlikely). As others have said, it’s hard to know what is the AI thought process or predict how it would react to a situation without testing it. So for all they know the bot could have said, in the first place, “Let’s die together.”

I haven’t used it nearly as much as VirtualBox but Boxes (flatpak) is definitely a breeze to use. It uses KVM under the hood I think. If your use cases are complicated it might abstract away too much though.

Not saying flaws make them useless, I’m saying the flaws mean they shouldn’t be a single point of failure.

The context size wouldn’t have really mattered because the bot was invested in the fantasy. I could just as easily see someone pouring their heart out to a bot about how they want to kill people but said in a tactful way that the bot just goes along with it an essentially encourages violence. Again, the bot won’t break character or make the connection that this isn’t just make believe, this could lead to real harm.

This whole, “It wasn’t me, it was the bot,” excuse is a variation on an excuse many capitalists have used before. They put out a product they know little about but they don’t think too hard because it sells. Then hundreds of people get cancer or poisoned and at worst there’s a fine but no real blame or jail time.

Character AI absolutely could create safeguards that would avoid harm but instead they’re putting in the maximum effort it seems to do nothing about it.