OpenAI shakes the world ofartificial intelligence. The company of Sam altman has decided to disband its team dedicated to risk mitigation long term of AI, known as “Superalignment team“. The group, established in July 2023, was created to evaluate the safety and ethics of using advanced AI. For a few weeks, however, OpenAi has begun to reintegrate team members into other projects and research. The dissolution of the AI safety team occurred following the resignation of co-founder Ilya Sutskever and the team co-leader Jan Leike in controversy with company top management.
Leike: “I thought OpenAI was the best place in the world”
Leike has expressed his dissent publicly regarding the company's priorities, saying that AI safety has taken a backseat to the production of more advanced models. In a post on X, he shared his concerns and disappointment at seeing safety culture overlooked. He emphasized the urgent need to understand how to guide and control AI systems that are more intelligent than humans.
“I joined OpenAI because I thought it was the best place in the world to conduct my research. However, I have been at odds with the company's leadership for a long time about what its priorities should be, and now we have reached the breaking point. Over the past few months, my team and I have gone against the grain in our own company. In recent years, the culture of security and ethical processes have been overshadowed by new products".
“OpenAI must become a company that puts security first,” Leike then wrote in a post on X, inviting all OpenAI employees to “act with the responsibility” necessary for what they are building.
OpenAI doesn't stop, start with the new ChatGPT and agreement with Reddit for training
In an interview with Bloomberg, OpenAI has dissolution of the team confirmed “Superalignment” AI safety strategy. The group has been disbanded as regulators' scrutiny grows. In the meantime, however, Altman's company does not stop and has recently launched a more advanced version of the chatbot, which incorporates multimodal modes, using not only text but also voice and video. Sam Altman compared this innovation to the artificial intelligence present in films, citing Scarlett Johansson's character in “Her Her”.
Furthermore, OpenAI has partnered with Reddit, the social news and forum platform, to use the platform's contents intraining artificial intelligence models like ChatGPT. Similar to a previous deal between Reddit and Google, this partnership will allow Reddit to offer AI-powered tools to users and moderators, while OpenAI will also become an advertising partner of the platform.
OpenAI uses large volumes of text to train its AI systems, often drawing from online data, including newspaper archives. The company has signed agreements with various newspapers, such as the Financial Times and the Associated Press, to gain access to their content, but there have also been legal problems: The New York Times and authors such as George RR Martin and John Grisham are suing OpenAI (and Microsoft) for copyright infringement in the unauthorized use of their materials.
Altman: no OpenAI browser at least for now
Sam altman he also denied Reuters' announcement a few days ago regarding the upcoming launch of a ChatGPT-based browser by OpenAI. He clarified the situation via a post on social networks, confirming that there will be neither GPT-5 nor a ChatGPT-based browser at the moment. At least for now. In fact his statements do not contradict completely the idea of an OpenAI browser with integrated ChatGPT, even if perhaps the project is just postponed to a later date.
