AI developers, particularly Anthropic and OpenAI, are encountering significant internal and external challenges as they push for a deceleration in AI development to prioritize safety. Internally, there are divisions, as evidenced by former Anthropic and OpenAI researcher Jacob Coxon's public resignation and warning about the companies racing towards "self-improving superintelligence." Coxon's post, which garnered 153 million views, highlighted a concern among some employees that the rapid pace of development is gambling with human lives, even though many within these companies are aware of the risks.
Externally, these calls for caution are clashing with powerful forces. Wall Street and the Trump administration are expected to resist any substantial slowdown in AI development. President Trump explicitly stated, "We’re leading China in AI… whoever wins AI, wins," downplaying safety warnings and emphasizing continued leadership. This puts pressure on AI companies, who are also grappling with competitive dynamics, as no single company wants to fall behind rivals, making a voluntary, widespread slowdown difficult to achieve without strong guarantees.
Despite the public calls for a slowdown by leaders like Anthropic CEO Dario Amodei and OpenAI's Sam Altman, and even support from Elon Musk, concrete actions to halt development have been limited. Amodei warned that unaligned AI could "take over the entire internet" within six to twelve months, leading to his three-part plan for pacing development. However, leading AI companies have not truly slowed down, instead focusing on more limited measures such as allowing independent safety experts access to their models and discussing the establishment of a new industry standards body. These discussions include OpenAI, Anthropic, and Google DeepMind coordinating on safety measures, with OpenAI's global policy chief Chris Lehane stating that these engagements have been ongoing for weeks.
