Consider two scenarios in which AI sector news scores an identical overall sentiment of moderately positive. In the first, the coverage is dominated by earnings beats, product launches, and bullish analyst commentary. In the second, the same score emerges from a mixture of modestly positive financial results and a growing number of articles about regulatory investigations and geopolitical risk, with the positive stories just edging out the negative ones. The aggregate number is the same. The situations are entirely different.
Narrative frame distribution is Canary’s way of capturing this distinction. Rather than simply asking whether articles are positive or negative, it asks what type of story each article is telling. This classification layer, introduced in early 2026, has proved to be one of the system’s most valuable outputs.
The Eight Frames
Every article processed by Canary is assigned to one of eight mutually exclusive narrative frames by Anthropic’s language model. The frames are designed to cover the principal ways in which AI sector news is structured, with each representing a qualitatively distinct type of story.
Revenue beats, expansion into new markets, bullish analyst upgrades, record shipments, and positive guidance revisions. This frame dominates during periods when the sector is performing strongly and confidence is high.
New product launches, model releases, chip architecture advances, capability milestones, and research with clear commercial implications. This frame reflects the pace of innovation in the sector.
Quarterly earnings, guidance announcements, analyst ratings changes, valuations, and mergers and acquisitions. This is the most routine frame in the cycle of AI sector coverage.
Government investigations, antitrust proceedings, export controls, data privacy enforcement, and proposed legislation affecting the sector. The growing share of this frame is often an early signal of a changing risk environment.
Trade tensions between major economies, China and United States friction in particular, supply chain disruption linked to international relations, sanctions, and foreign government restrictions on AI deployment.
Market share losses, new entrants disrupting incumbents, pricing pressure compressing margins, and notable customer defections. This frame reflects the internal competitive dynamics of the sector.
Share price declines, selloffs, missed earnings expectations, guidance cuts, layoff announcements, and asset writedowns. The rising share of this frame often lags negative developments but confirms their breadth.
External forces acting on the sector, including interest rate policy, broader market conditions, energy costs affecting data centre economics, and global economic uncertainty where the AI industry is present but the larger story lies outside it.
Reading the Distribution Chart
The Narrative Frame Distribution chart on the dashboard is a stacked bar chart showing, for each day, the proportion of articles falling into each of the eight frames. The bars always sum to one hundred per cent, so what the chart reveals is the changing balance of story types over time rather than absolute volumes.
On a typical stable day, Growth Momentum and Financial Results will tend to dominate, reflecting the ordinary cadence of AI sector reporting. What draws the analyst’s attention is when this balance shifts. When Regulatory Risk and Market Correction frames begin to take a larger share, or when Technical Breakthrough coverage declines despite a market that appears, on sentiment scores alone, to be holding up, the distribution chart is often the first place the change is visible.
This predictive quality is not accidental. Research across multiple studies of financial news behaviour confirms that categorising news by the type of story being told is a better predictor of market volatility than sentiment scoring alone. The frame distribution is capturing something that valence cannot: the structural character of the news environment, not just its emotional temperature.
Continuing and Reversal: Tracking Frame Momentum
Canary’s frame analysis does not simply report today’s distribution in isolation. Each frame is compared against its equivalent in the prior week, and the dashboard displays whether the direction of any shift is continuing from the previous day or represents a reversal.
The practical significance of CONTINUING labels becomes clear when several risk frames — Regulatory Risk and Market Correction for example — show CONTINUING movement simultaneously over multiple days. This pattern represents something qualitatively different from day-to-day noise. It reflects a genuine rotation in the type of news being published, which typically precedes a corresponding rotation in sentiment scores.
Research into how financial markets respond to communication has found that markets calibrate to stability: when a sustained period of consistent messaging gives way to a significant change, the response tends to be larger than it would have been without the preceding period of calm. A series of consecutive CONTINUING labels establishes an expectation; a subsequent reversal of that trend can therefore carry more signal weight than a single isolated change would suggest.
What to Watch For
The most actionable patterns in the frame distribution chart are not single-day readings but sustained directional shifts. A single day’s elevated Regulatory Risk share might reflect one prominent story and fade immediately. The same frame showing CONTINUING movement for three or more consecutive days, particularly if accompanied by a declining share in Growth Momentum or Technical Breakthrough, is a different category of signal entirely.
These sustained frame rotations are what the Semantic Volatility Index is specifically designed to quantify and track over time. The frame distribution chart provides the qualitative picture; the Semantic Volatility Index provides the composite number.