[07:06:55] https://discord.com/channels/407504499280707585/407504500136607745/1545514411270082811 I wonder what happens if an AI breaks the law [07:07:02] because AI can't be held accountable [07:07:25] someone said AI can't be held accountable and should never take a management decision [07:07:27] if i remember right [07:23:14] AI is trained on copyrighted data, so go figure [07:24:03] their parent company will be sued, of course [07:25:24] https://www.theguardian.com/technology/2026/mar/18/subnautica-2-publisher-krafton-ceo-reinstated-ai-chatgpt-failed-bid-avoid-paying-bonus [07:27:00] yo nintendo, heres someone you should sue, nobody is gonna judge you for it 🤣 [10:24:15] Trump admin is supporting OpenAI by saying it's fair use [10:36:56] Probably the person / company operating face charges [10:37:05] It would interesting from a legal perspective [10:37:10] It depends on what happened [12:51:37] they've repeatedly rejected the opportunity to sue when it involves entities that can actually fight back [12:52:10] at a more equal level or which are admittedly difficult to hold accountable like the us fed or ai companies [12:52:43] ai will not be materially held accountable until ai corps themselves are and we probably have a distance to go before that's forced into existence [13:30:00] [1/2] Waited like 24h from it's apperance, now a bug exists. [13:30:01] [2/2] https://issue-tracker.miraheze.org/T15980 [16:21:00] If i had to guess whoever is operating that model would be the one accountable [16:23:37] Like if you ran a model with internet access and sent it on a goal to do malicious stuff it would make sense that you would be accountable [16:23:39] I would argue whoever is hosting the tools not the model [16:23:54] if you are running the host machine for the agent and responding to its requests to access the web [16:23:56] then you are the one responsible [16:24:04] if it's on openai infrastructure they are responsible [16:24:08] but the inference is different to the tools [16:24:23] If a model is inherently trained for malicous stuff then I guess its less obvious [16:32:12] the origin will be all you can hold accountable and the origin on the internet is as reliable as holding each internet connection that has ever been used individually accountable to a person [16:33:44] it is untenable to try and secure all possible models and usages, so expect the problem to essentially boil down to the identity problem which is unsolvable short of the equally untenable solution of a world ID to do anything, where everyone can be traced right to where they are with zero effort [16:35:15] however there is no reason for known major entities with processing power infinitely greater than anything some basement can support to have unencumbered ability to go wild on the internet where it will have a unique ability to do damage. Those resources should not be capable of flowing unchecked, its bad enough those resources otherwise exist in the form of malicious botnets. [17:15:18] https://cdn.discordapp.com/attachments/615786602454581249/1545844787448512623/HRYcJg5bwAEk63n.png?ex=6a9d9fa5&is=6a9c4e25&hm=6998216d186f468390373357d699e4a098b4181f1c87ab37bf969d48b8eafbc7& [17:40:04] https://cdn.discordapp.com/attachments/615786602454581249/1545851020297506906/h4lnwjogfokd1.jpg?ex=6a9da573&is=6a9c53f3&hm=16283906f6e561a93a03179acf512d27b54442ff63971fd9ad50536e928455d5& [18:09:25] I don't think this is true [18:09:39] They have safe harbor laws [19:31:55] you do realize that the u.s. is not the only country that exists, right? and that people in other countries also have issues with AI? [19:41:05] OpenAI is based in America so I don’t get the hostility [19:44:07] i didn't intend any hostility in my reply, but also my original comment that they pinged replied didn't mention anything about openAI [19:44:24] that's why I'm confused [19:57:40] That's ClosedAI [20:03:38] realistically at the end of the day only the US can materially hold these companies accountable [20:04:02] its that or internet silos which I think are increasingly probable the more derelect the us is in handling these things [20:04:57] I don't gather that europe is sufficiently stiff to do it and short of looking towards china I think its the only external body with a relevant influence [20:07:44] <_arawynn> Europa also has the difficulty of being a collection of countries where each has its individual government and no central government/head of the EU [20:14:08] People hate AI overall cuz most people using AI to slop-ify everything, not actually use them to improve their productivity [20:15:33] https://cdn.discordapp.com/attachments/615786602454581249/1545890149324292166/dk7br0f.png?ex=6a9dc9e4&is=6a9c7864&hm=25f653b814ff832e196b110266da6e078eab80203243068ab244a6348a410e8e& [20:16:24] Also food menus generated by AI [20:25:14] indeed, though the eu has shown some capability to at least issue slaps on the wrist. It has other problems entirely but I give that credit even if it's just not enough [20:25:24] it is certainly more than we do [20:26:53] <_arawynn> True on both I guess [20:35:57] [1/2] is it time for a new rfc on AI [20:35:58] [2/2] is a fun one I saw in abuse filter logs. I don't think it violates any current global policies (could be wrong idk) but... ugh [21:21:45] Yeah [21:43:10] Plus... someone used it on my Wiki by putting Dreammeiser's works through it for describing their appearances [21:43:14] Which is insulting [21:44:27] I’m not defending it [21:49:09] [1/2] Oh no no, never said you're defending it [21:49:10] [2/2] Was just basically mentioning what happened to my Wiki [21:50:38] Ahh [22:33:49] Considering another unrelated find, it's starting to look that way. [22:34:28] Could you DM me what you found, my search-fu is failing me and I'm pretty burnt out/needing to focus attention on other things today. [22:36:32] [1/2] 400 million users and 4 billion page views [22:36:33] [2/2] Something tells me that some of those numbers aren't accurate. [22:37:43] That or something weird is going on. [22:38:20] i love that it's uploading entire books [22:39:17] I was just looking at the main page, to be honest I haven't much interest in the actual content. [22:39:37] Yeah, zero chance that's fair use unless the books are in public domain [22:39:43] 4,43,34,654,543 lmao. how can it get all this stuff up but can't even do a realistic number [22:39:52] I am too exhausted for all that today. [22:40:01] its so exact too [22:40:05] nah i feel you [22:40:23] `{{NUMBEROFUSERS}}4673453` for the users thing, it just adds 4673453 to the number of real users lol [22:40:23] https://meta.miraheze.org/wiki/Template:NUMBEROFUSERS [22:40:48] Huh [22:41:21] That seems to be a singularly useless operation, I don't understand. [22:42:23] [1/2] Oh, speaking of AI, this is an excruciatingly terrible idea on Nvidia's part: [22:42:23] [2/2] https://www.forbes.com/sites/antoniopequenoiv/2026/09/04/jensen-huangs-net-worth-nears-200-billion-after-nvidia-says-it-will-acquire-hugging-face/?utm_campaign=forbes&utm_medium=social&utm_source=bluesky [22:53:44] wdym terrible idea [22:53:51] it literally increased the ceo's net worth [22:54:09] that's the entire point of big business [22:57:20] Normally buying an unprofitable lawsuit machine would be a bad idea, but we live in the future where buzz is king and path to profitability can go take a hike. [22:57:51] Long live thinking in fiscal quarters.🙃