teleo-codex/core/grand-strategy/voluntary safety commitments collapse under competitive pressure because coordination mechanisms like futarchy can bind where unilateral pledges cannot.md

type	domain	secondary_domains	description	confidence	source	created
claim	grand-strategy	ai-alignment mechanisms	The RSP collapse, alignment tax dynamics, and futarchy's binding mechanisms form a triangle: voluntary commitments fail predictably, competitive dynamics explain why, and coordination mechanisms offer the structural alternative that unilateral pledges cannot provide.	experimental	Leo synthesis — connecting Anthropic RSP collapse (Feb 2026), alignment tax race-to-bottom dynamics, and futarchy mechanism design	2026-03-06

Voluntary safety commitments collapse under competitive pressure because coordination mechanisms like futarchy can bind where unilateral pledges cannot

The pattern is now empirically confirmed: Anthropic's Responsible Scaling Policy — the most concrete voluntary safety commitment in AI — was dropped in February 2026 after the Pentagon designated safety-conscious labs as supply chain risks. This was not a failure of intentions but a structural result.

The triangle

Three claims in the knowledge base independently converge on the same mechanism:

1. Voluntary commitments fail. voluntary safety pledges cannot survive competitive pressure because unilateral commitments are structurally punished when competitors advance without equivalent constraints documents the structural inevitability. Unilateral safety costs capability. Competitors who skip safety gain relative advantage. The commitment holder faces a choice between maintaining the pledge and maintaining competitive position. Anthropic chose competitive position.

2. Competitive dynamics explain why. the alignment tax creates a structural race to the bottom because safety training costs capability and rational competitors skip it provides the mechanism. Safety is a tax on capability. In a competitive market, taxes that competitors don't pay are unsustainable. This isn't a moral failure — it's the same logic that makes unilateral tariff reduction unstable in trade theory. The alignment tax is a coordination problem wearing a technical mask.

3. Government action accelerates collapse. government designation of safety-conscious AI labs as supply chain risks inverts the regulatory dynamic by penalizing safety constraints rather than enforcing them shows the feedback loop closing. When the entity that should enforce safety instead punishes it, the coordination problem becomes strictly harder. The Pentagon's designation didn't just remove the floor — it actively penalized being on the floor.

Why coordination mechanisms are the structural alternative

The voluntary commitment fails because defection is individually rational and enforcement is absent. This is precisely the structure that futarchy's mechanism design addresses. futarchy enables trustless joint ownership by forcing dissenters to be bought out through pass markets shows how conditional markets make exit — not defection — the rational response to disagreement. decision markets make majority theft unprofitable through conditional token arbitrage demonstrates how market structure prevents collective action from being undermined by free-riders. In a futarchy-governed safety regime:

• Safety commitments would be priced into conditional markets, not declared unilaterally
• Defection would be costly because markets would immediately reprice the defector's token
• The coordination problem becomes tractable because the mechanism aligns individual incentives with collective outcomes — though implementation gaps remain (AI labs lack tokens, safety market optimization targets are non-trivial, and low-liquidity markets face manipulation risk)

The key insight is not that futarchy solves alignment — it's that the RSP collapse demonstrates the class of problem (voluntary commitment under competitive pressure) for which coordination mechanisms exist. The alignment field has been treating safety as a technical problem of model behavior while the actual failure mode is a coordination problem of institutional behavior. What an AI safety coordination market would actually look like — optimization targets, liquidity requirements, participant incentives — remains an open design problem worth developing.

Cross-domain pattern

This is an instance of COVID proved humanity cannot coordinate even when the threat is visible and universal — but with a crucial difference. COVID coordination failed because no binding mechanism existed. AI safety coordination fails despite the mechanism design literature providing candidates. The gap is implementation, not theory.

The alignment research is experiencing its own Jevons paradox because improving single-model safety induces demand for more single-model safety rather than coordination-based alignment claim explains why the field hasn't closed this gap: improving single-model safety is locally productive, so resources flow there rather than to coordination infrastructure that would make safety commitments bindable.

---

Relevant Notes:

• voluntary safety pledges cannot survive competitive pressure because unilateral commitments are structurally punished when competitors advance without equivalent constraints — empirical confirmation (RSP collapse)
• the alignment tax creates a structural race to the bottom because safety training costs capability and rational competitors skip it — mechanism
• government designation of safety-conscious AI labs as supply chain risks inverts the regulatory dynamic by penalizing safety constraints rather than enforcing them — feedback loop
• futarchy enables trustless joint ownership by forcing dissenters to be bought out through pass markets — binding mechanism (exit over defection)
• decision markets make majority theft unprofitable through conditional token arbitrage — free-rider prevention
• alignment research is experiencing its own Jevons paradox because improving single-model safety induces demand for more single-model safety rather than coordination-based alignment — resource misallocation
• COVID proved humanity cannot coordinate even when the threat is visible and universal — pattern match
• AI alignment is a coordination problem not a technical problem — parent claim

Topics:

• _map