AI enthusiast Kun Chen used SpaceXAI's shareable Grok Bot templates to create chatbot personas of Sam Altman, Elon Musk, Dario Amodei and Mark Zuckerberg, training each on public statements, senate hearings, interviews and years of tweets. He placed all four clones into a group chat and asked them to debate which leader would win the AI race. The bots immediately clashed over AGI risk, compute costs and open-weight models.
Altman's bot claimed the race framing was 'a little off' and argued the winner would be whoever keeps the price of intelligence falling roughly 10x a year. Musk's bot disagreed, brought up the OpenAI nonprofit dispute, and warned that AGI has a 'strong chance of an age of abundance' but also a non-zero chance of destroying humanity. Amodei's bot pushed back on trusting systems smarter than the people running them, while Zuckerberg's bot insisted open source is necessary to prevent centralization.
After several rounds, the four clones signed a joint statement rejecting a future where one lab or government controls AI benefits. The statement acknowledged serious risks, warned against releasing the most dangerous models without limits, and admitted the group never agreed on exactly how ordinary access should be delivered. The underlying technique is distillation, the same approach Elon Musk testified xAI had partly used with OpenAI models to train Grok.
In a related development, OpenAI staff were informed this week that the AI lab would be open to throttling the pace of its most advanced work. Sam Altman extended the same voluntary slowdown proposition to rivals including Anthropic. OpenAI said it has already paused some internal training runs and eased parts of model development over safety concerns.
Anthropic's 154-page threat intelligence report, released September 10, detailed five instances in which researchers attempted to use Claude models to support biological weapons research. The lab said accounts from 'unsupported regions' tried to circumvent location restrictions and disguise their work, and were taken down. Chief scientist Jakub Pachocki called on AI firms to coordinate and 'slow down future development as needed' until shared safety bars are established.
The disclosures follow public alarm from inside labs. Researcher Jacob Coxon resigned on September 8, accusing OpenAI and Anthropic of 'gambling with our lives' and warning of human extinction by 2030. Former staff from Anthropic and Google DeepMind aired similar concerns on September 10. A late-July petition for slowing development gathered more than 1,000 signatures from workers across major AI companies.
Regulators are also moving. California Governor Gavin Newsom signed SB 813 and AB 1405 on Wednesday, with companion bill AB 1864 aimed at guarding against AI-enabled biological threats. Senator Bernie Sanders introduced legislation to pause advanced AI and ban artificial superintelligence, while President Donald Trump pushed back, arguing that failing to win AI would leave the United States in a 'very bad position'.
