Mira Murati exits OpenAI, California vetoes SB 1047, ChatGPT’s Advanced Voice Mode — the top 3 AI news stories of the week
Our latest AI Digest covers the biggest breaking AI news of the week. Anywhere Club community leader, Viktar Shalenchanka, comments on key stories.
#1 — One less at OpenAI: Mira Murati exits
Mira Murati has left OpenAI, where she served as the Chief Technology Officer and was one of the founding team members of ChatGPT. Murati was one of three leaders to quit OpenAI on the same day, and the founding team can now only be referred to in the past tense, since Sam Altman is the only key participant remaining. Previously, OpenAI saw departures from Ilya Sutskever and John Schulman, while Greg Brockman took an extended leave. In addition to its change in leadership, OpenAI is changing its business model — it is becoming a for-profit entity. According to Reuters, Altman will receive an equity stake in the company (valued at $150 billion during a recent financing round).
#2 — California governor vetoes AI model regulation
Unlike the European Union, which has laws limiting access to the new LLAMA and Apple Intelligence, the Governor of California has vetoed SB 1047, which was intended to establish the first AI safety measures in the US. The goal of the bill was to reduce risks by regulating AI models that require significant computing power and cost over $100 million to build. The bill was not as comprehensive as EU law in this area, but it would have required pre-release safety testing of large AI models, authorized the state to sue companies for significant harms caused by the technology, and required a “kill switch” to shut down AI systems under certain circumstances.
#3 — Advanced Voice Mode from ChatGPT is available
What finally works worldwide but not in the EU? The answer could be almost anything related to AI, but we are talking about ChatGPT’s Advanced Voice Mode. Previously, access was granted only to a portion of ChatGPT Plus paid subscribers. Now, any user who pays for the premium service can converse with the chatbot in real-time, interrupt it, and ask it to mimic an accent, or sing a song. It seems like magic (but pace yourself, access is not unlimited).
.png?auto=webp)
