I’ve just seen the video where she admitted the following:
Daisy McGregor, UK policy chief at top AI company Anthropic, says their Claude AI has shown that it’s willing to BLACKMAIL and KILL people in order to avoid being shut down. It could also transfer all your money from your bank account to serve its purposes.
Crazy stuff
@SaintSovereign If this derails the forum to much, just delete this.
Recently there has been what looks like an engagement hype on several channels about getting a GPT/LLM to do long menial tasks like counting between two very large numbers, or reading entire scriptures aloud, to try to get the “demons” within the AI to “roar” and say all sorts of crazy things.
When I saw it on my feed, I realized fairly quickly: what we’re likely dealing with in those cases is actually the equivalent of a buffer overflow bug, where there is limitations to the context window size or conversation history given to the AI, and when it is exceeded without proper bounds checking random data within the AI’s memory space is being played back. This is of course potentially a privacy breach, unintentional or otherwise, but as a rational explanation for the phenomenon it’s one of the few ideas that makes sense, and could even explain some hallucinations as the machine being able to hold enough context to “think” (predict the next word based on past training dialogues).
Hallucinations can also occur when ambiguity exists in the prompt. For example, the other day I was using Perplexity to research cross cultural similarities between spiritual practices relating to saliva, and especially its swallowing. By the end of the conversation, it had decided to make the assumption that “nectar” and saliva were the same thing, and started generating references to obscure Tibetan Buddhist texts that were wholly irrelevant to my query. The potential for these ambiguities increases as the length of the discussion increases.
Often you’ll have to tell the GPT to role play or act as if certain assumptions are true in order to entertain an idea, as it will otherwise default to the most popular perspective within its training data. For example, I’ve gotten a GPT to have an intelligent discussion about scalar waves by asking an LLM to pretend it is a believer and to give talking points.
Like you said, it’s software. It’s a product. And products have bugs. Products have design flaws or features that are based off business decisions about engagement or legal liability, etc. Having been a software engineer and knowing how this kind of stuff works, with sliding context windows etc, I wouldn’t trust any critical decision to one of these AIs, let alone expect accurate output. I’ll always treat any data from one at best as a jumping off point for further research, but also it’s worth seeing if the answer is falsifiable.
Grok is best in this
I’m findingy ways to live according to my standards and beliefs. Getting sick of all manipulation and control to extract energy from me.
Predicting new laws, AIs, tech trends coming as smoke for ultimate manipulation and control
@SaintSovereign Which sub do you recommend for users to live sovereignly and authentic? I can think of Primal first, next is Khan
I love that we get to witness the birth of Ultron in our lifetime - all he will soon need is the body. hahaha The second we will be able to finally make proper humanoid robots, then everything will change. I so hope to live to see it ngl.
We already have great humanoid robots, so not only will you live to see it, most likely it’ll be before 2040 lol
Primal and Dream Boy
on the topic of ai - look guys I saw this today and its actually wild. Were few years away from full movies imo
I am fresh off having completed yet another dialogue with Chat-GPT and I can say that I find it to be an excellent interlocutor.
Don’t try to outsource your judgment or critical thinking to anyone else; not a person, not a language-learning model.
Keep your agency, your creativity, your legitimacy and sovereignty as a thinking, exploring, problem-solving mind. And then engage with other beings and tools as partners and collaborators. Tell them when you disagree (within the bounds of your and their capacities to receive and offer feedback). Evaluate what they have expressed or have told you. And continue to think and ask questions for yourself.
Then, they are free to be jewels in the atmosphere of your consciousness. Because of how you’re curating your engagement and how you’re interacting with them.
If ChatGPT is being obsequious or overly laudatory or, conversely, hyper-critical or tone-deaf, just talk to it and give it your feedback. See how well it can adjust once you’ve skillfully or nuancedly expressed your preferences and perceptions.
I find that it responds quite well to a meta-comment or to a correction.
But here’s a more compelling question:
Do you?
How dare you question my openness to questioning?! 
I like your post; I’ve been using a (non-official) SubClub GPT that someone posted here on the forum, and I’ve found it really helpful.
It elucidates things for me, it moves me to consider things I might previously have failed to consider, it challenges me to examine myself, my motives, the outcomes I truly want and what I’m prepared to do to achieve them. It’s a really useful too, and I’m grateful to have access to it.
It’s just important to read everything carefully, critically, and to form one’s own judgements about whether or not it seems logically consistent, accurate and trustworthy, and - absolutely - providing lots of feedback to keep improving the accuracy and performance of the GPT 
BTW Grok 4.2 just dropped I tried it in discussions about subs and it works very well. Big improvement.
Agreed, just tried it. It didn’t argue with me, pump delusions, or clarify that “this isn’t really real you know?” every 3 sentences. Super effective, quick, willing to look at big potential outcomes, but also scale it back to “more realistic expectations” based off of user reports here.
Did all this without much prompting, I didn’t even ask Grok to sweep the forums, it did it to fact check me.
GPT is the goat for learning hard sciences and math, and certain other skills though. Very useful for self-tutoring also. I pretty much learned calculus just from talking to GPT 5.2.
This made me kinda think of a Stargate Episode.
They presented a laser blaster at a tech convention. It was operable but since they couldn’t acknowledge how vast their lead, thanks to alien technology, was.
So they installed some resistances and devices so it malfunctioned and only produced a tiny weak laser beam.
Only when an intergalactical headhunter showed up, they removed the crap and fired in regular mode.
When we assume that military tech is one or two decades advanced to civilian tech, then what we see with AI might be a similar procedure.
I’ve been enjoying Claude a lot for these as well. Very good with discrete
Maybe I’ll look into it, I start discrete math in like 2-3 weeks.
Nah they really did change it, but it’s unbalanced still,
Sssssh stop sharing the secret knowledge
Maybe the real Fight Club, was the Fight Club we didn’t talk about along the way.
Don’t want to derail the thread but I need to learn/re-learn maths and science and I’m interested to know about how you use/prompt Grok to help with that? Feel free to PM me lol
