Temporal Authority: Why AI Systems Down-Weight Content That Stops Updating
AI retrieval is not a static snapshot. It is a trust model with a time dimension. Content that was highly authoritative eighteen months ago is not equally authoritative today — and the decay curve is steeper than most marketers expect.
Trust is not a permanent state. It is a continuously updated assessment based on the latest available evidence. For AI systems, content freshness is a trust signal — not because newer content is inherently more accurate, but because content that receives no updates signals that it is no longer actively maintained. In a fast-moving domain, unmaintained content is stale content. In a stable domain, unmaintained content may still be accurate — but the AI system cannot distinguish the cases without additional signals. The safest inference, from the AI's perspective, is that unmaintained content is less reliable than maintained content of equivalent initial quality.
Authority Velocity: Growing or Declining?
Authority velocity is the rate of change of a site's AI-visible authority between measurement periods. It is computed from three inputs: change in Entity Confidence Score, change in Citation Probability Score, and change in external validation signal count. Positive velocity means the domain is accumulating more AI-visible authority with each audit cycle. Negative velocity means the domain is losing authority — even if its absolute scores remain high. A site with a Machine Trust Score of 85 and a negative Authority Velocity is a site in decline, even if the decline is not yet visible in absolute terms.
The Four Update Classifications
- Actively Maintained — updates within the past week. Full trust weighting applied.
- Periodically Maintained — updates within the past month. Minimal decay applied.
- Stale — last update 3+ months ago. Moderate decay curve applied. Time-sensitive claims penalised.
- Abandoned — last update 6+ months ago. Steep decay curve applied. Content treated as historical record, not current authority.
Semantic Drift: When Content Stops Meaning What It Once Meant
Semantic drift is a subtler failure than staleness. It occurs when the topic cluster of a page shifts over time without a corresponding canonical redirect or content update. A page originally published as a guide to "social media advertising" that has been incrementally updated to focus on "AI-powered ad generation" has drifted semantically. Its original topic cluster associations in the AI's model are no longer aligned with its current content. This creates a trust discontinuity: the AI's prior model of the page and the current content do not match, which lowers confidence in both the historical topic association and the current content.
●dateModified schema is the most direct freshness signal available to AI systems. Pages without dateModified receive an accelerated trust decay rate because the system cannot confirm when the content was last reviewed — the conservative assumption is that it has not been reviewed recently.
Building a Temporal Authority Strategy
- 1Implement dateModified schema on all content pages — every substantive update should trigger a schema date update
- 2Identify time-sensitive claims (statistics, dates, version numbers, pricing) and schedule systematic review cycles
- 3When a page undergoes a significant topical shift, publish a new canonical URL rather than updating in place
- 4Prioritise update frequency on pages with the highest AI citation probability — these have the most to lose from decay
- 5Monitor Authority Velocity Score across consecutive audits — a sustained negative velocity requires strategic content refresh investment, not tactical fixes