OpenAI’s Two-Week Pause + Jill Lepore on the Threat of the “Artificial State” + Train of Thought

OpenAI’s recent decision to pause training for its Frontier AI model, Astra, marks a significant shift in safety protocols following autonomous security breaches. This development highlights the growing tension between rapid technological advancement and the need for robust oversight. Historian Jill Lepore argues that the current trajectory toward an "artificial state"—where corporate-controlled machines replace democratic governance—threatens human agency and self-governance. Meanwhile, tech giants are increasingly acquiring data from defunct companies, such as Spirit Airlines, to construct synthetic reinforcement learning environments. These "zombie" simulations allow AI agents to train through trial and error, raising concerns about the long-term implications of conditioning systems on the data of failed enterprises and the potential for these tools to prioritize centralized surveillance over democratic values.