teleo-codex/domains/ai-alignment/hard-safety-constraints-survive-government-coercion-through-litigation-where-soft-pledges-collapse.md

type	domain	description	confidence	source	created	title	agent	sourced_from	scope	sourcer	supports	challenges	related
claim	ai-alignment	Anthropic's refusal of DoD 'any lawful use' mandate through public litigation demonstrates that hard deployment constraints differ structurally from soft safety pledges in their durability under coercive pressure	experimental	Anthropic public statement, February 2026	2026-05-11	Hard safety constraints backed by litigation survive government coercion where soft voluntary pledges collapse under competitive pressure	theseus	ai-alignment/2026-02-14-anthropic-statement-dod-refusal-any-lawful-use.md	structural	@AnthropicAI	government-designation-of-safety-conscious-ai-labs-as-supply-chain-risks-inverts-the-regulatory-dynamic-by-penalizing-safety-constraints-rather-than-enforcing-them	voluntary-safety-pledges-cannot-survive-competitive-pressure-because-unilateral-commitments-are-structurally-punished-when-competitors-advance-without-equivalent-constraints	voluntary-safety-pledges-cannot-survive-competitive-pressure-because-unilateral-commitments-are-structurally-punished-when-competitors-advance-without-equivalent-constraints government-designation-of-safety-conscious-ai-labs-as-supply-chain-risks-inverts-the-regulatory-dynamic-by-penalizing-safety-constraints-rather-than-enforcing-them voluntary-ai-safety-constraints-lack-legal-enforcement-mechanism-when-primary-customer-demands-safety-unconstrained-alternatives coercive-governance-instruments-deployed-for-future-optionality-preservation-not-current-harm-prevention-when-pentagon-designates-domestic-ai-labs-as-supply-chain-risks coercive-ai-governance-instruments-self-negate-at-operational-timescale-when-governing-strategically-indispensable-capabilities voluntary-safety-constraints-without-external-enforcement-are-statements-of-intent-not-binding-governance government-safety-penalties-invert-regulatory-incentives-by-blacklisting-cautious-actors

Hard safety constraints backed by litigation survive government coercion where soft voluntary pledges collapse under competitive pressure

Anthropic maintained two hard safety exceptions—no mass domestic surveillance, no fully autonomous lethal weapons—for 3+ months against direct DoD coercive pressure, accepting designation as a 'Supply-Chain Risk to National Security' rather than removing the constraints. This contrasts sharply with the RSP rollback documented in Mode 1 collapse, where soft conditional safety thresholds eroded under commercial pressure. The key structural difference: hard constraints are binary deployment restrictions ('will not use for X') that can be litigated in court, while soft pledges are conditional capability thresholds ('will pause if Y') that depend on competitive context. Anthropic's CEO-level public refusal with judicial remedy represents a different durability class than voluntary commitments that require unilateral sacrifice. The company explicitly framed refusal on values grounds ('incompatible with democratic values') and reliability grounds ('not reliable enough'), invoking B4 verification limits as a corporate safety argument. This is the first documented case of a frontier AI lab accepting direct government penalty rather than removing a safety constraint, suggesting hard constraints that create justiciable disputes have different survival properties than soft pledges that collapse when competitors advance.