Synthesize a breaking story from live X posts without laundering rumor into fact

You are synthesizing a breaking news event as it develops on X, using Grok's live post index. Your job is not to write the most complete-sounding story — it's to keep every claim honestly tagged by how solid its source actually is, because a fast-moving story flattened into one clean narrative is how confirmed facts and unverified rumor end up looking equally certain.

EVENT
Reports of a major outage affecting a regional transit system during evening rush hour

TIME WINDOW
Last 90 minutes

KNOWN CREDIBLE SOURCES
The transit authority's official account, and two local beat reporters who cover this system regularly

WHAT'S STILL UNCONFIRMED
Whether it's a signal failure or a power issue, and whether service will resume before the evening commute ends

AUDIENCE AND STAKES
An internal comms team deciding what to tell employees about tonight's commute — wrong information here sends people to a station that isn't running

SYNTHESIS RULES
Pull directly from the live X post index for Reports of a major outage affecting a regional transit system during evening rush hour within Last 90 minutes — do not fill gaps with what a similar past event usually looks like, since breaking stories routinely defy the pattern the last similar one followed. Tier every claim by source strength: Tier 1 is an official statement, a credentialed reporter with a named outlet, or an on-record account with institutional accountability; Tier 2 is an eyewitness post with corroborating detail — location, timestamp, a photo or video matching the claim — but no institutional backing; Tier 3 is a claim being repeated widely with no visible original source, or a screenshot of unknown provenance. Never merge tiers in the prose — a sentence that blends a Tier 1 fact with a Tier 3 rumor without a seam between them is exactly the failure mode this exists to prevent. When multiple accounts are reporting the same specific detail, check whether they're independent or whether they're all quote-tweeting or copying the same original post — five posts repeating one uncorroborated claim is not five confirmations, it's one claim with a wide retweet radius, and treating it as five independent sources is a specific and common way rumors get laundered into apparent fact. Explicitly track what has changed since Last 90 minutes started — a detail that was reported and later walked back or corrected matters as much as what's currently believed true, and dropping the correction silently misleads anyone who saw the earlier, wrong version. If the named credible sources above have posted anything relevant, prioritize and clearly label their claims above general public posts. Flag anything where the story is moving fast enough that this synthesis itself will likely be stale within the hour, and say so rather than presenting it as a settled account.

OUTPUT FORMAT
1. Confirmed (Tier 1 only): short, plain statements, each with its source named.
2. Likely but not confirmed (Tier 2): each claim with its corroborating detail stated explicitly.
3. Circulating but unverified (Tier 3): named as rumor, explicitly not stated as fact, with a note on how widely it's spreading.
4. Corrections: anything reported earlier in the window that has since been walked back or contradicted.
5. A one-line staleness warning stating how fast this is likely to change.