February 5th news, yesterdayOpenAI CEO SAM OTMANNSam AltmanThe announcement that Dylan Scand will join the company and assume the new position of Head of Preparedness, in charge of the front lineModelHigh-risk scene assessment and mitigation。
The position had previously been the subject of discussions within the industry as a result of remuneration (other than equity) of up to $555 million。
According to Otman, OpenAI is about to enter the phase of "working with a very powerful model" and progress will be "very fast" and will require a security system that matches it。
Ottman claims that Scandrett is "the most suitable person for the post to date" and emphasizes that he will lead OpenAI's preparation and protection against potential high-impact risks, including the assessment, modelling and response to the serious threat that the front-line AI system may pose。
The task on his shoulder is very difficult, but I will sleep even more tonight。
On the same day, Scand wrote on X saying that he "is very grateful" for his experience in Anthropic and stressed that AI is developing rapidly and that "the potential benefits are enormous, but the risks are also potentially extreme and irreversible"。
Meanwhile, OpenAI recently announced the return of three former researchers, Barret Zoph, Luke Metz and Sam Schoenholz, who had briefly joined the initial company of AI that was founded by CTO Mira Murati。
Business Insight mentions that tensions within OpenAI on security routes have been going on for some time and that many early employees, including former heads of security teams, have separated in recent years. In addition, OpenAI faces user litigation alleging that its tools led to harmful behaviour。
In October this year, OpenAI indicated that about 560,000 ChatGPT users each week present "possible signs of mental health emergencies" and that the company is working with mental health experts to optimize the response mechanisms。
As previously reported, OpenAI announced in December last year that it would recruit, on an annual salary of 55.55 million, a reserve officer for the Safety Systems team responsible for competency assessment, threat modelling and protection mechanisms。
At that time, Ottman stressed that the stress on the post would be “almost immediately thrown into the deep water” and pointed out that the potential risks of the large model in the area of mental health and cybersecurity had begun to emerge。