The Siren Song of the Inside Game
The author argues that the AI safety community disproportionately overinvests in the inside game of working closely with frontier labs compared to maintaining an adversarial outside posture. This imbalance is driven by structural factors such as resource concentration, talent density, personal risk aversion, and institutional incentives. The piece suggests that rebalancing these strategies through intentional resource allocation and subsidized independent criticism is increasingly necessary. Open article →