Whistleblower behaupten, OpenAI habe seinen Mitarbeitern illegal untersagt, Sicherheitsrisiken offenzulegen
https://www.washingtonpost.com/technology/2024/07/13/openai-safety-risks-whistleblower-sec/
Whistleblower behaupten, OpenAI habe seinen Mitarbeitern illegal untersagt, Sicherheitsrisiken offenzulegen
https://www.washingtonpost.com/technology/2024/07/13/openai-safety-risks-whistleblower-sec/
4 Kommentare
Submission statement: how should society protect whistleblowers as we get closer and closer to AGI?
At what point should it go from being like regular whistleblowers to something different?
What security events have happened and will happen that we might not know about because nobody felt they could safely let the public know?
Odd, I only skimmed the article but I didn’t see any explanation of what the specific whistleblower reports were. Isn’t the whole point of whistleblowing to make that information available?
They’re not safety risks, because there is no risk to safety from an LLM. All these ’safety‘ people do is censor the models for prudes.
What they barred them from is releasing trade secrets and badmouthing the company to negatively affect their investment.
There are no safety risks.
This whole thing is so shockingly stupid.
That they notified the SEC first is highly informative. This isn’t a problem of dangerous tech, it’s a problem of bullshit hype.
Don’t get me wrong, technology can be dangerous – bad code has resulted in people being irradiated by medical devices – but AI „risk“ is being hyped as a way to make people think regulating AI is necessary so the big players csn keep small upstarts out of the market.