There has been a feud between OpenAI CEO Sam Altman and Anthropic CEO Dario Amodi for over 10 years.
Dario Amodi, CEO of Anthropic, the AI company that develops Claude AI, worked at OpenAI, the AI company that operates ChatGPT and other applications, from 2016, but left OpenAI in 2020 due to differences in business direction. The Wall Street Journal has compiled a report based on interviews with sources familiar with the matter, stating that there has been a feud between CEO Amodi and OpenAI CEO Sam Altman for about 10 years.
The Decadelong Feud Shaping the Future of AI - WSJ
Anthropic was in contract negotiations with the U.S. Department of Defense, but the Department of Defense pressured Anthropic to remove its established safety measures. CEO Amodi refused, stating that he would not succumb to threats. As a result, President Donald Trump ordered federal agencies to cease using Anthropic's technology and designated the company as a supply chain risk. Meanwhile, OpenAI reached an agreement while maintaining essentially the same safety standards as those Anthropic had rejected.
OpenAI reaches agreement with Department of Defense on AI usage contract, shortly after negotiations with Anthropic fall through - GIGAZINE
In signing the contract with the Department of Defense, OpenAI stated that it had 'imposed almost the same restrictions as Anthropic,' and that it had made clear agreements such as 'not allowing any use of AI technology for surveillance of the American public.' However, CEO Amodi reportedly wrote in an email to employees that 'the main reason OpenAI accepted the deal with the Department of Defense and we did not was that OpenAI was interested in appeasing Department of Defense personnel, while we were actually interested in preventing abuse,' and criticized OpenAI's claim that it had imposed restrictions as 'a blatant lie.'
Anthropic CEO Dario Amodi points out that 'OpenAI's message is a blatant lie' regarding the contract with the Department of Defense - GIGAZINE
by Fortune Brainstorm Tech
In the first place, CEO Amodi left OpenAI due to differences in business direction, and since founding Anthropic in 2021, he has positioned his company as a 'healthy alternative to AI competitors.' For this reason, he has emphasized safe and responsible AI development, such as by launching a campaign criticizing OpenAI for integrating advertisements into its chatbots. However, it is said that there was a deep-seated animosity between CEO Amodi and CEO Altman even before Amodi and others left OpenAI due to concerns that OpenAI did not prioritize safety.
Keach Hazy of the Wall Street Journal, who interviewed more than 250 people, including CEO Altman and those around him, and wrote the book ' The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future,' points out that OpenAI co-founder Greg Brockmann is strongly involved in the conflict between CEO Amodi and OpenAI executives.
According to Hazy, the conflict dates back to 2016. At the time, CEO Amodi, who was working as an AI researcher at Google, was living in a shared apartment with his sister Daniela Amodi and her fiancé, Holden Karnovsky. Brockmann was a frequent visitor, and one day, while discussing 'the right way to develop AI,' Brockmann argued that 'the details should be shared with the American public,' while CEO Amodi insisted that 'sensitive topics like the speed of AI development should be communicated to the government first.' Brockmann later recalled that 'it was an incident that symbolized the difference in thinking between OpenAI and Anthropic.'
CEO Amodi joined OpenAI in 2016, but in 2017, Elon Musk, who was funding OpenAI at the time, demanded that a list of every employee's contribution be compiled, and subsequently, about 10-20% of the staff were laid off. Amodi says he felt this was 'unnecessarily cruel' and developed distrust towards the company. Also in the same year, a concept was discussed in which 'OpenAI would take on the role of coordinating AI systems for governments around the world in the future,' and Brockmann proposed the idea of 'selling artificial general intelligence (AGI) to governments around the world.' However, Amodi believes that 'selling AGI to hostile countries is tantamount to an act of treason' and even considered resigning.
When Musk left OpenAI in 2018, CEO Altman became the de facto leader. CEO Amodi agreed with Altman to remain at OpenAI on the condition that 'Brockmann and Ilya Satskebaa would not be in charge.' However, Brockmann later stated that 'we have been told that we can dismiss CEO Altman if we think it is necessary,' and CEO Amodi felt that CEO Altman was not keeping his promise.
The conflict deepened further when development of the GPT series began. CEO Amodi, who was then in charge of research, opposed Brockmann's involvement in the development of a large-scale language model. CEO Altman tried to persuade CEO Amodi to allow Brockmann to participate in the project, but Daniela also insisted that she would resign from her position if Brockmann was allowed to participate, and ultimately Brockmann did not participate in the project.
CEO Amodi increased his presence within OpenAI by leading the development of GPT-2 and GPT-3. However, he felt that his contributions were not being properly recognized, for example, because 'Brockmann, who contributed less to the company, was giving opinions to CEO Altman on important company policies.' He was also furious when he was asked by Brockmann to review presentation materials, only to later learn that the materials were for a meeting with former President Barack Obama, and that he had been excluded from the meeting.
In 2019, CEO Amodi was promoted to Vice President of Research, but the internal conflict did not subside. In 2020, CEO Altman alleged that the Amodi siblings were gathering criticism against him at the board meeting, and when Amodi and Daniela denied it, a heated argument ensued. Furthermore, after executives conducted mutual evaluations, Brockmann and Daniela exchanged lengthy criticisms of each other, and the conflict within the organization became irreparable.
At the end of 2020, a group led by CEO Amodi decided to leave OpenAI. CEO Altman tried to persuade them to stay, but Amodi demanded 'direct reporting to the board of directors' and 'not being able to work with Brockmann,' and negotiations broke down. Ultimately, Amodi, Daniela, and about 10 others left OpenAI and founded Anthropic. Currently, Anthropic is one of OpenAI's biggest competitors.
The following image shows a scene from the '
Related Posts:
in AI, Posted by log1e_dh