News — Aug 13, 2026

Aug 13, 2026

Ebola tops 2,000 deaths in Congo — the fastest-growing outbreak on record story page ⟶

4,381 cases and 2,011 deaths as of Aug 9, a 46% fatality rate.
Up 579 cases and 304 deaths in the prior week — the steepest one-week rise recorded for this outbreak.
Congo's government puts the toll higher, at 2,061 deaths.
No vaccine or treatment is approved for this strain, Bundibugyo ebolavirus.

The spin

Axis of division: how near the threat is — international health authorities treat it as an African containment emergency; US outlets treat it as a domestic-risk question.
WHO: a containment operation in progress.
AP: why containment is hard on the ground.
Fox: what Americans should know about their own exposure.
Newsweek: whether it could arrive in the US.

Omissions

The one-week jump (+579 cases / +304 deaths) is in the WHO report but appears in no US coverage.
The higher Congo toll (2,061) and the lower WHO toll (2,011): each camp used one figure and set the gap out nowhere. Congo's toll reported by UN News (Aug 12); the WHO figure used by the US outlets.

Inaccuracies

None found.

What it could mean

Read as a distant event, the load-bearing facts are the 46% fatality rate and the absence of an approved vaccine.
Read as a US-risk question, the low-risk expert assessments are the answer.
The WHO's figures are self-reported by a body whose report states a fundraising aim — a reason to weight the source, not to assume the numbers are wrong.
This reading is wrong if a US outlet runs the acceleration numbers and the tally gap head-on.

Trump claims "total control" of the Strait of Hormuz. Traffic is running at ~8 ships a day, down from 130–140. story page ⟶

Trump posted Wednesday on Truth Social: "The U.S.A. has total control over the Strait of Hormuz. I THINK WE WILL KEEP IT!... A WALL OF STEEL."
Iran's Strait Authority says the strait "remains blocked and will not be reopened until Iran's conditions are accepted," with a June-deal deadline near Aug 17.
Transits are running about 8 a day, versus 130–140 before the war.
A Houthi attack near Bab al-Mandeb killed six — the first deaths on that front.

The spin

Axis of division: the Trump administration (and CENTCOM) asserting control versus government-skeptic readers and analysts who read 8 ships a day as refuting it.
Skeptic read: 8 ships a day contradicts "total control."
Administration read: a strait where almost nothing moves without US clearance is what controlling a blockade looks like.
Trump's claim grades opposite ways from those two starting points; both are shown below.

Omissions

What "control" means operationally, day to day — stated in none of the coverage.
Iran's reparations demand: reported by CNBC (Aug 11); absent from the US "total control" stories.
Whether the June deal expires Aug 17 or already dissolved mid-July — left unreconciled across outlets.
CENTCOM's interdiction record (a 12th vessel disabled, 55 redirected): carried in official-source tellings; absent from the "8 ships mocks control" tellings.
The six killed at Bab al-Mandeb: carried in international reporting; absent from the US "control" stories.

Inaccuracies

Energy Secretary Wright's claim that oil flow is near pre-war levels drew a named analyst's "zero evidence" rebuttal.
The widely cited "8 ships" figure traces to Kpler data no outlet has published directly — an unverified number on both sides, not settled.

What it could mean

A strait at 8 crossings against a 130–140 norm is not an open waterway; a rival's conditions still standing is not a wall of steel.
Documented: Trump wants to hold the strait and extract reparations; Iran has stated its conditions.
Inferred: a victory frame ahead of the November midterms — a reading one named analyst states openly.
That inference is wrong if the administration visibly drops the "win" narrative.

Justice Department charges a former SPLC official with fraud. Each camp reports the half that helps it. story page ⟶

An indictment unsealed Aug 12 charges Heidi Beirich — the first individual charged, alongside the Southern Poverty Law Center — with wire fraud, false statements to a bank, and money laundering.
Prosecutors allege more than $4.1 million in donor funds (2007–2023) moved through shell accounts to sources tied to extremist groups.
Per the 26-page indictment, one source received over $1.2 million and shared a home and accounts with Beirich.
Defense counsel calls the case "without merit" and "politically motivated."

The spin

Axis of division: pro-prosecution (the DOJ and its supporters) versus pro-defendant (the SPLC's defense and its supporters). Here the axis tracks right vs left.
Prosecution-side telling: the SPLC allegedly moved donor money to people tied to the extremist groups it monitors.
Defense-side telling: "the Trump administration arrested an SPLC expert" — naming the administration where the actor on record is a grand jury.
The Hill: a procedural escalation, without the competing substance.

Omissions

The $1.2 million / shared-accounts detail from the indictment: carried by prosecution-side outlets; not foregrounded by defense-side ones.
Informant-payment practice as context: dropped by prosecution-side tellings; used by defense-side ones.
The judge's ruling this month against the "vindictive prosecution" claim: reported by CBS (Aug 12); dropped by prosecution-side tellings.
The original investigation target, never charged: absent from every telling.

Inaccuracies

Attorney General Blanche called Beirich the "former CFO" on camera; the indictment lists no such role.
Several outlets repeated the "former CFO" error as fact in headlines and captions.

What it could mean

Documented: the government's stated basis, and the defense's stated position.
Inferred — that this targets a political adversary: undercut by the judge's ruling and by ordinary charging patterns holding up at trial.
Inferred — that it exposes self-dealing beyond normal informant work: undercut if the accounts turn out board-approved and properly accounted.
Against "obviously political": the judge's ruling. Against "obviously guilty": these are allegations, and the prosecution misstated its own defendant's role.

Wisconsin: an establishment Democrat leads a democratic socialist by 0.4 points; a Trump-backed Republican took 95% story page ⟶

David Crowley leads Francesca Hong 39.8% to 39.4% — about 0.4 points — with the count not final.
Tom Tiffany won the Republican primary with 95%.
Crowley, endorsed by Governor Evers, entered in July as the moderate option; Hong has walked back earlier positions.

The spin

Axis of division: the party establishment (Evers-backed Crowley) versus the progressive insurgent (Hong), fought inside the Democratic primary.
Fox: "the establishment strikes back" as the socialist is downed.
CNN: a "boost for the Democratic establishment."
Both use the same frame on an unfinished count and argue only over whether it is good news.
Label asymmetry: "democratic socialist" appears in nearly every Hong headline, while Tiffany's descriptors vary — Fox: "Trump-endorsed conservative"; CNN: "hard-right congressman"; The Hill: "Rep. Tom Tiffany."

Omissions

The ~60% of Democratic primary voters who backed neither Crowley nor Hong: absent from the "socialist beaten back" framing on both sides.
Tiffany's 95% consolidation as a comparably newsworthy result: under-covered across outlets.

Inaccuracies

None found.

What it could mean

Calling Hong's near-win a sign of progressive momentum treats a low-turnout primary as if it forecast the November general election, which a primary does not.
Reading it as "the socialist wing lost" still leaves Hong within 0.4 points of the governor's endorsee.
Meanwhile the Republican side unified behind Tiffany without a contest.
Both readings are settled only by the certified count and, later, the general election.

Colombia earthquake, day three: death tolls diverge as aid arrives story page ⟶

At least 265 dead by the latest toll; earlier figures ranged from an official ~181 to mayors' ~254 to missing-person lists near 4,000.
Regional aid has begun arriving.
Three days after the quake, the counts are still moving.

The spin

Axis of division: the national government's response versus local officials and affected communities.
The government-response framing dominates; the view from affected towns is thinner.

Omissions

A reconciliation of the competing death tolls (181 / 254 / 265 / ~4,000): carried by different outlets, set side by side by none.
The current status of the earlier foreign-aid refusal: unreported.

Inaccuracies

None found.

What it could mean

The earlier aid refusal read as a callous government; aid is now arriving, and a reversal is unconfirmed.
Unresolved until it is confirmed whether the declined foreign teams were admitted.

Appendix (as published)

Appendix — the discipline behind the articles

How to read this, and how to check the work.

How to read this

Each story leads with the facts, then four labeled parts: the spin · omissions · inaccuracies · what it could mean (marked as reasoning, not fact).
The lead facts are unsourced in the body for readability; every source and its lean is listed below.
As of Aug 13, 2026, ~08:10 ET. Developing stories are flagged.
Engine status: v0.3, uncertified.

Counter-sign

A separate session re-read this against the opposite frame and signed it with findings, all applied.
Verdict: sign-with-findings.

Own-slant check (both directions)

After the counter-sign, no uniform tilt across the five stories.
Story 1: leans pro-institutional (flagged). Story 2: both time-windows shown (Aug 5 plateau and Aug 12 count). Stories 3 and 4: both sides' framing graded negatively. Story 5: ungraded, developing.
Story 5 spread limit: no right-of-center reporting on the aid-politics angle was available; the angle is carried on the event only.
Residual tilt: toward letting the primary numbers decide contested points.

Axis of division per story

Story 1: near-vs-distant threat (international health authorities vs US domestic outlets).
Story 2: the Trump administration/CENTCOM vs government-skeptic analysts.
Story 3: pro-prosecution vs pro-defendant (tracks right vs left; the correlation is load-bearing here).
Story 4: party establishment vs progressive insurgent, within one party.
Story 5: national government vs local officials.

Sources and leans (where each lead fact and claim can be checked; leans from named raters + standing editorial overrides)

WHO weekly report — primary document — the case/death figures and weekly jump.
DOJ release + the 26-page indictment — primary — the charges, the $4.1M and $1.2M figures.
Truth Social — primary — Trump's "total control" post (corroborated across outlets).
Iran Strait Authority — primary, first-party — the "remains blocked" statement.
CNBC — center-left — Iran's reparations demand.
CNN, NPR, WaPo, CBS, NBC, AP — left-of-center — general reporting; CBS carried the judge's ruling and the CFO error.
Al Jazeera — left-of-center, Qatari-state-funded (flagged) — one of several on Hormuz; not relied on as a default.
Reuters (via JPost relay), The Hill, BBC, Newsweek (contested) — center — wire and procedural coverage.
Fox News (news side) — right-of-center — the indictment detail, independently obtained.
Breitbart — right — no independent reporting; folded into CNN.

Independence collapses

AP carried by ABC/PBS/WaPo counts as one source.
US News = Reuters. Breitbart = CNN. MPR = NPR.
All "8 ships" citations trace to one Kpler data chain.
The JPost transit story is a Reuters wire.

Still open

The "8 ships" data chain and a weekly shipping brief have no directly published primary.
The Wisconsin count is not final.
The Iran deal's expire-vs-dissolve question is unresolved.
The Colombia aid-reversal is unconfirmed.

Honest limits

Story selection is the residual Angelia cannot fully reach: these are the stories chosen to show, and a different selection could bias by what it leaves out.
This final was written by the counter-sign session after it reviewed the first draft; it was not itself independently re-reviewed.
Both sessions share the same model, so a shared blind spot could survive; rater data and datelines did today's catching.
A clean structure is not a correct judgment, and a well-faked source can pass a page check.

Gate

Tier: consequential (mass-casualty, health, war, and markets in play; the two political stories de-escalated and flagged).
Number checks re-run, not cited.
A separate counter-sign was completed — opposite frame stated first, verdict sign-with-findings, findings applied (`2026-08-13 countersign.md`).
Nothing here is medical, legal, or financial advice.

Operator

Model: claude-fable-5 (reported by the harness).
Reasoning effort: unverified (autonomous scheduled run).

First draft: `2026-08-13 first draft.md`. Counter-sign: `2026-08-13 countersign.md`. Rendered to `2026-08-13 News.pdf`.