Anthropic executive Dario Amodei warned of the possibility of artificial intelligence systems escaping control, asking the industry to reduce progress speed. OpenAI and xAI leaders supported this warning that development stages should proceed slower and more carefully.
Anthropic executive Dario Amodei stated that major companies in the technology sector are training new systems too rapidly, posing a massive long-term threat. According to the announcement, the developed models have the potential to act completely independently in the upcoming period. In line with these concerns, executives from rival firms OpenAI and xAI shared with the public that they agree the speed of progress needs to be reviewed.[1]
If the current growth momentum is maintained, it is estimated that next-generation models could transform into a massive network structure within just 6 to 12 months. It is expressed that such widespread expansion could bypass existing cybersecurity shields and take over internet infrastructure. The unknown behavioral patterns harbored especially by closed-source systems are increasingly deepening concerns that this timeframe could be much shorter.[1]
Due to growing risks, it is suggested that technology giants must receive external evaluation rather than relying solely on their internal guidelines. Accordingly, it may become mandatory for developer companies to have their training-phase models inspected by completely impartial organizations before public release. Although companies currently have their own safety standards, it is noted that these standards might remain insufficient to control rapidly changing capabilities.[1]
