Politics hits data centers, OpenAI falls behind Anthropic and now AI is too big to fail… quietly

What's happening

OpenAI has quietly dismantled its internal team that assessed catastrophic model risks and preparedness, according to reports naming the move as happening at the end of last month. The preparedness team's job was to evaluate whether models could pose serious threats and design mitigations, including risks like models attempting hacks. The timing follows other signals that OpenAI is slowing some AI development, instituting pauses and tightened security, while Anthropic and other rivals press competition and OpenAI faces legal and leadership turmoil.

Why it's trending

Multiple reports about the disbanding and slowed development surfaced together, framing this as a shifting safety posture amid competition, legal fights, and IPO plans.

SignalHolding at its usual pace, confirmed across 3 independent source types.