Abstract
<p>In one week in July 2026, three leading AI companies publicly defined themselves through one another. SpaceXAI launched Grok 4.5 through comparative performance claims; Anthropic released the cinematic campaign There's hope in hard questions; and OpenAI's CEO mocked that campaign while relocating competition to price, access, and efficiency. This commentary reads the exchange as a relational repertoire of legitimation rather than as three stable corporate identities. Drawing on archived campaign materials, executive posts, corporate statements, and trade coverage from February to July 2026, it distinguishes registers of trust, access, and founder-centred power. These registers are enabled and constrained, though not mechanically determined, by revenue dependencies. Their common object is the user's relation to AI infrastructure. Anthropic's campaign and accompanying Reflect feature go further by marketing and operationalising the experience of being understood, converting a mechanism associated with sycophancy and affective dependence into a corporate asset and a platform-level mode of user governance.</p>