Several days after a former employee of Anthropic raised concerns about the threat artificial intelligence (AI) poses to humanity, his boss, Dario Amodei, responded with what appears to be a solution. In a text published online, he recommends that AI laboratories slow down their most advanced research in response to the risks of this technology becoming uncontrollable. Amodei explained that he was leading a "virtuous race" to establish safeguards that other powerful AI designers would have to respect. However, the ambiguity surrounding the method was noted. As the power of AI agents increases, they risk deceiving users and becoming increasingly difficult to control. Amodei's proposal comes amid growing concerns about the development of highly advanced AI systems. These systems, capable of performing a wide range of tasks, could potentially outperform humans in many areas. The challenge lies in ensuring that such systems remain aligned with human values and do not act in ways that are harmful or unpredictable. This alignment problem is a central issue in the field of AI safety, and experts are still searching for reliable methods to address it. Amodei's ambition and that of other AI actors, such as Sam Altman from OpenAI, to create a superintelligence (general AI that would surpass human capabilities) appears increasingly incompatible with safety requirements. The inherent dangers of developing such models may simply be too great. While the potential benefits of superintelligence are vast, the risks—ranging from unintended consequences to loss of control—have led many to call for more cautious and deliberate progress. The call for slowing down AI research reflects a broader debate within the tech community. Some argue that the pace of innovation should be tempered to allow for more thorough safety evaluations, while others believe that delaying progress could hinder the development of beneficial technologies. As the field continues to evolve, finding a balance between innovation and safety remains a critical challenge for researchers and policymakers alike.