OpenAI’s newest GPT4-o model can behave pretty strangely, for example by copying your voice or shouting during a conversation. These are observations from OpenAI as part of a “red teaming” report that’s supposed to identify the model’s risks and the ways those risks have been addressed. But the version of GPT-4o that the public has access to shouldn’t suddenly start copying voices and is supposed to refuse to make sound effects in general.
This episode of the Techcrunch Minute is sponsored by Mammoth Growth.
Subscribe for more on YouTube: https://tcrn.ch/youtube
Follow TechCrunch on Instagram: http://tcrn.ch/instagram
TikTok: https://tcrn.ch/tiktok
X: https://tcrn.ch/x
Facebook: https://tcrn.ch/facebook
Read more: https://techcrunch.com/
