The OpenAI series, the father of ChatGPT, continues again this weekend. After Ilya Sutskever, the second head of the security team, Jan Leike, in turn announced his departure.
There is a lot of movement within the American start-up, which has just announced an impressive new development with GPT-4o. We learned a few days ago of the departure of the man who tried to oust Sam Altman from his post, namely Ilya Sutskever. A departure which will not seem to be lonely, since Jan Leike, who was the other major security manager, in turn communicates on the end of his work within OpenAI.
Security Team Reaches Final Stage of Disbandment
We didn’t know it, but for several weeks, OpenAI had been carrying out a real reorganization of its services. The so-called “superalignment” team, which was responsible for overseeing the values that an artificial general intelligence (AGI) – AI capable of performing all the intellectual tasks of a human being – should adopt – was in the process of being disbanded. This was illustrated by the departure first of Ilya Sutskever, then of Jan Leike, who produced a thread on X to detail the reasons for his resignation.
He thus explained that in his opinion, “ In recent years, security culture and processes have taken a back seat to flashy products. » He also indicates in the same thread that it has been increasingly difficult for this team to obtain resources to develop security means. This is why he wanted to finally leave his job, while explaining that OpenAI had to get its act together on this essential issue. “ OpenAI takes on a huge responsibility on behalf of all humanity » he wrote.
OpenAI must realize its responsibilities according to Jan Leike
So even after he resigned, he made a point of advising the company on what it should do. “ OpenAI must become a security-focused AGI (artificial general intelligence) company » he points out. And this while it seems that Microsoft’s partner has already internally brought into the world an incredibly powerful AI, called Q*.
Jan Leike finally completed what looks like a professional testament with a direct message, this time to OpenAI employees. “ Learn to feel the AGI. Act with the gravity that suits what you are building […] I count on you » he urged them.
A thread that has not left the major players in the industry indifferent. Sam Altman for his part greeted the former colleague, indicating that Jan Leike “ right, we still have a lot to do [sur la question de la sécurité]. » Elon Musk responded directly to her thread, saying that “ therefore, security is not a top priority at OpenAI. »
Download
- Chat in different languages, including French
- Generate, translate and obtain a text summary
- Generate, optimize and correct code
Created by OpenAI, ChatGPT is an advanced chatbot powered by the latest generation GPT-4 language model. By leveraging deep learning and artificial intelligence technologies, this chatbot has the ability to decipher and understand user requests. Thanks to its ability to generate text in an ingenious way, ChatGPT offers tailored and relevant responses, ensuring smooth chat interaction and an optimized user experience.
Created by OpenAI, ChatGPT is an advanced chatbot powered by the latest generation GPT-4 language model. By leveraging deep learning and artificial intelligence technologies, this chatbot has the ability to decipher and understand user requests. Thanks to its ability to generate text in an ingenious way, ChatGPT offers tailored and relevant responses, ensuring smooth chat interaction and an optimized user experience.
Source : The Verge, Le Figaro
0