Post

Jesse Felder
@jessefelder.com

‘This company simply cannot be trusted. Their software is becoming ever more dangerous; their internal security practices leave a lot to be desired; they aren’t being straight with the public; and their proxies are preparing us to swallow the damage...’ garymarcus.substack.com/p/pause-open...

88 reposts214 likes

Tilly "Tots”
@tiltnweave.bsky.social

Talk to me like I'm a dumb teenager -- what are we talking about? Hijacking bank accounts? Taking over government websites? Stealing personal information? We need to know *what* to fear - -why this is so scary.

D-Cym
@d-cym.bsky.social

The CEO of OpenAI is more concerned with maximizing his personal profit and wealth expansion than making sure the product he is foisting on the world is safe. That is the threat. Safety is being ignored to maximize profit.

Tilly "Tots”
@tiltnweave.bsky.social

And, you missed my point. People need to know, more specifically, what is the "threat"? Safety against what? If you can't tell people what they need to be concerned about, they won't listen or care.

D-Cym
@d-cym.bsky.social

Accessing the computer systems of other people/companys is a crime. What part of multiple AI's conspiring in committing a crime which the CEO said they couldn't do bc of safety protocols which weren't implemented, and then the CEO downplaying the seriousness of the probelm leaves you unconcerned?

MarkMan
@madmanmark84.bsky.social

Amen. Amanpour had a good interview with Heidy Khlaaf the other day, which explained (in fairly "lawman terms") what I said a month ago, how OpenAI has been trying to spin their *human* security incompetence & irresponsibility into positve marketing PR, "our product is super powerful!"

pbs.orgWhat Really Happened When OpenAI Bots Escaped a Cybersecurity Test?When OpenAI set out to test how well its product could hack, hundreds of AI agents broke out of their test environment and infiltrated the AI platform Hugging Face. Headlines warned that the machines ...
MarkMan@madmanmark84.bsky.social

Correct. It's complicated, but TL;DR: OpenAI is obfuscating that their *humans* weren't careful enough. OpenAI was negligent in properly sandboxing the model - internet access was limited only through configuration. Thinking it was "caged," OpenAI then INTENTIONALLY GAVE IT NEFARIOUS INSTRUCTIONS