Rich Harang
@rich.harang.org
Using bad guys to catch math since 2010.
Principal Security Architect (AI/ML) and AI Red Team at NVIDIA.
He/him. Personal account etc; `from std_disclaimers import *`
Safe AI starts with Secure AI.
Principal Security Architect (AI/ML) and AI Red Team at NVIDIA.
He/him. Personal account etc; `from std_disclaimers import *`
Safe AI starts with Secure AI.
I do! Think the talk went well, thanks!
You around for DC too?
You around for DC too?
August 8, 2025 at 12:51 AM
I do! Think the talk went well, thanks!
You around for DC too?
You around for DC too?
Goodhardt's law undefeated etc etc
June 11, 2025 at 11:34 PM
Goodhardt's law undefeated etc etc
And let's be honest: capitalism.
At some point, engagement farming + person-specific fine-tuning + microtargeted advertising will wind up with someone's free online waifu knowing exactly how to sell them dropshipped Temu crap in bulk in a way they're not mentally prepared to defend against.
At some point, engagement farming + person-specific fine-tuning + microtargeted advertising will wind up with someone's free online waifu knowing exactly how to sell them dropshipped Temu crap in bulk in a way they're not mentally prepared to defend against.
May 1, 2025 at 11:52 AM
And let's be honest: capitalism.
At some point, engagement farming + person-specific fine-tuning + microtargeted advertising will wind up with someone's free online waifu knowing exactly how to sell them dropshipped Temu crap in bulk in a way they're not mentally prepared to defend against.
At some point, engagement farming + person-specific fine-tuning + microtargeted advertising will wind up with someone's free online waifu knowing exactly how to sell them dropshipped Temu crap in bulk in a way they're not mentally prepared to defend against.
We've already seen chatbots convince people who were mildly predisposed to mental health problems talk them into self-harm or trying to kill public figures. This seems likely to only get worse and impact broader and broader segments of the population as the engagement tuning gets better.
May 1, 2025 at 11:50 AM
We've already seen chatbots convince people who were mildly predisposed to mental health problems talk them into self-harm or trying to kill public figures. This seems likely to only get worse and impact broader and broader segments of the population as the engagement tuning gets better.