Risk, Crisis and Emergency Communication Weekly

Coverage period: 08:00 on 27 July 2026 to 07:59 on 3 August 2026, Europe/Lisbon time. This episode includes only material first published or released in that window. It is written for experienced communication professionals and separates proposals, official accounts, independent reporting, preliminary findings and evidence gaps.

This week in one paragraph

The week’s strongest signal is that communication systems are being judged at the point where people act, not where a message leaves an institution. The US is tightening the security and proposed design of emergency alerts. Australia’s AusAlert test shows why cell coverage is only one measure of warning performance. WHO’s Ebola and mpox reporting puts community trust, cross-border coordination and visible data cut-offs together. Two new misinformation studies and reporting from China show the limits of one-size-fits-all countermeasures as AI-generated evidence enters disaster information flows.

Three developments to prioritise

1. Alerting is becoming a security, governance and accessibility problem

On 31 July, the Federal Communications Commission published two related documents. The final rule requires US Emergency Alert System participants to change default passwords, patch equipment promptly, use firewalls or network segmentation, restrict remote access and keep alerting equipment off the public internet. The rule is effective from 29 September. 1

The companion Further Notice of Proposed Rulemaking asks for comment on message authentication, duplicate suppression, more accurate geotargeting, possible retransmission rules, standardised threat symbols and text-to-speech for earthquake alerts. It is a proposal, not an operating rule, and the underlying FCC action was adopted in June. The July publication opens the formal comment period. 2

Australia offers the practical counterpoint. NEMA reported that the 27 July AusAlert test reached 94 per cent of mobile cells, above its 90 per cent target. It recorded no public-safety incident attributed to the test, no material change in driver behaviour and no material increase in calls to Triple Zero. Those are preliminary system indicators, not measures of comprehension or safe action. 3

ABC reporting added the harder cases: remote and satellite-dependent users, Queensland’s decision not to use AusAlert this bushfire season, and domestic-abuse charities warning that a loud test could reveal a hidden safe phone. Coverage is a cell metric; safety is a human and social outcome. 4

Fiji’s National Common Alerting Protocol workshop, reported by UNDP on 28 July, aims to create one alert and disseminate it through SMS, cell broadcast, radio, television, websites, apps and social platforms, with clear institutional ownership. Thailand’s disaster department made the same logic operational on 2 August, naming cell broadcast, social media, public address, community radio, village leaders and public-relations offices as part of flood preparedness. 5 6

Operational test: measure receipt, audibility, comprehension, recall, accessibility, safe response and fallback. Include old devices, remote areas, people who cannot safely expose a phone and people who receive advice through community intermediaries.

2. Outbreak communication must preserve trust across uncertainty and borders

WHO’s 1 August Disease Outbreak News reported 3,605 confirmed Bundibugyo virus disease cases and 1,587 deaths in the Democratic Republic of the Congo as of 30 July, across 49 health zones in five provinces. WHO described it as the largest Ebola outbreak recorded in the DRC. The figures are a snapshot, and the crude case-fatality ratio is not a forecast for an individual. 7

The communication point is in the response design. WHO says insecurity, displacement and access constraints are complicating containment, and stresses community-centred work led by local authorities and trusted community leaders. Uganda declared its smaller outbreak over on 28 July, but cross-border risk remains. A reassuring declaration in one country therefore needs a vigilance message in its neighbours, without turning vigilance into panic.

WHO’s 68th external situation report on mpox, published on 31 July, shows why data age belongs in the message. Its global data run to 30 June, African data to 19 July and operational updates to 22 July. It also names risk communication and community engagement, infodemic management and community-based infection prevention as part of the community-protection cluster. 8

Operational test: put a data timestamp in the headline or first sentence; separate confirmed facts, working hypotheses and unresolved questions; name trusted local messengers; and create a cross-border handover for rumours, case definitions and action advice.

3. AI is weakening the evidence chain around emergencies

BBC Verify reported on 31 July that Google paused an AI image tool integrated into Google Earth after tests showed that users could create false scenes on top of real satellite and three-dimensional map imagery. Invisible marks often worked, but detection tools and the AI assistant could still be misled. 9

BBC reporting from China on 27 July described fabricated flood videos, false emergency details and fake images of animals during real storms. It quoted an expert who argued that governments and platforms should prepare before false material goes viral, because detection technology is lagging behind generation technology. 10

Two pieces in the Harvard Kennedy School Misinformation Review add evidence, with important limits. A survey of 54 experts, published on 29 July, found that video deepfakes were judged the largest political threat, while generated text ranked highest in the health domain. The sample was small, purposive, heavily Western and included no respondents from Africa or Latin America. These are perceptions, not observed effects. 11

A systematic review published on 30 July compared professional and community fact-checking across 21 empirical studies. Professional systems tended to do better on trust; community systems tended to do better on scale. Neither performed strongly across trust, scalability and impact, and the literature was concentrated on X and Community Notes-style systems. 12

Operational test: include a false satellite image, a real image with a false caption and a genuine image that a detector misclassifies. Test who can verify it, how quickly a correction reaches the same audience and what the organisation says when verification is incomplete. Provenance, trusted messengers, platform escalation, human review and a correction channel need to work together.

Cases and evidence

Japan: dual hazards and an information gap

Reuters reported that a magnitude seven point one earthquake struck Kumamoto on 28 July, with about 300,000 people told to move to evacuation centres. Officials warned about aftershocks and landslides while also warning evacuees about heatstroke in temperatures around 34 degrees Celsius. NPR reported a death toll of 34 on 31 July. 13 14

The same reporting exposed a familiar gap: authorities did not yet have a consolidated missing-person figure. A dated knowledge timeline is useful here. Say what is confirmed, what is being reconciled, what families should do now and when the next update will arrive. Silence invites speculation; false precision makes correction look like concealment.

US water utilities: an act-now public warning

On 30 July, the FBI and EPA warned that utilities in at least seven states had reported unauthorised activity targeting internet-facing programmable logic controllers. The agencies recommended disconnecting exposed controllers, changing default passwords, auditing remote access, patching firmware, monitoring for changes and reporting incidents. 15

This is a useful model for sector-wide risk communication: it names the threat without attributing an actor it cannot verify, and it gives operators a short list of actions. The customer-facing question is separate: how will a utility explain service disruption without exposing operational vulnerabilities or creating unnecessary alarm?

Regional signals: Greece and Peru

The Associated Press reported that two firefighting helicopters collided near Athens during a wider European wildfire emergency. CNN reported that Peru suspended an airline after a Nazca Lines crash that killed 13 people, while family notification and technical investigation continued. The Peruvian cause remained unconfirmed, and the airline statement was reported through a secondary source, so these are regional signals rather than settled case studies. 16 17

Their shared lesson is sequencing: evacuation and safety advice first, condolence and family support alongside it, and technical or legal conclusions only when the evidence supports them.

What to test next

Put ownership on the alert. Who can approve it, authenticate it, translate it, correct it and speak when agencies disagree?
Maintain four evidence labels. Use confirmed, working hypothesis, unresolved question and next verification step across cyber incidents, outbreaks, evacuation zones and disputed images.
Measure correction performance. Count who saw the correction, whether trusted intermediaries repeated it, whether the original claim kept circulating and whether the correction changed the decision that mattered.
Write the fallback and stop conditions. If the network is down, what is next? If an image cannot be authenticated, who pauses publication? If a model produces an unsupported claim, who rejects it?

Watchlist

Australia’s AusAlert feedback and October launch preparation, especially device compatibility, the Queensland decision and safe-phone implications.
The FCC comment period, which closes on 31 August, and whether emergency communicators submit evidence on accessibility and end-to-end performance rather than technical reach alone.
DRC Ebola figures, the 18 August International Health Regulations review and WHO’s next mpox update.
Follow-up reporting from Kumamoto on dual-hazard messaging and the missing-person information gap.

Editorial transparency

This edition is limited to the stated seven-day window. Older studies promoted during the week, undated social posts, vendor material and communication-light items were rejected. The main evidence gaps are first-party research or guidance from Latin America and the Caribbean, the Middle East and South Asia, plus non-English primary pages. Those regions are represented through cases and operational signals only, not as proof of balanced publication output.

The episode was researched, selected and scripted with AI assistance. Human editorial review checked the date boundary, opened source pages, removed duplicate developments and kept the evidence limits audible. The narration is synthetic. A source claim remains a source claim.

Direct-source list

Federal Communications Commission, Federal Register. “Wireless Emergency Alerts; The Emergency Alert System; Modernization of the Nation's Alerting Systems.” Published 31 July 2026. Type: proposed rulemaking. Region: United States. Access: open. Direct source
Federal Communications Commission, Federal Register. “Modernization of the Nation's Alerting Systems; Protecting the Nation's Communications Systems From Cybersecurity Threats.” Published 31 July 2026. Type: final rule. Region: United States. Access: open. Direct source
National Emergency Management Agency. “AusAlert national test a success.” Written 31 July 2026. Type: official test result. Region: Australia. Access: open. Direct source
ABC News. “AusAlert test a success as millions of devices nationwide receive alert siren.” Published 27 July 2026. Type: independent reporting. Region: Australia. Access: open. Direct source
United Nations Development Programme, Pacific Office. “Fiji Advances Multi-Hazard Early Warning Systems Through National Dialogue on Common Alerting Protocol.” Published 28 July 2026. Type: official workshop report. Region: Fiji and the Pacific. Access: open. Direct source
The Nation Thailand. “Northern provinces step up flood preparations as heavy rain looms.” Published 2 August 2026. Type: reporting of official guidance. Region: Thailand and Southeast Asia. Access: open. Direct source
World Health Organization. “Ebola disease caused by Bundibugyo virus — Democratic Republic of the Congo.” Published 1 August 2026. Type: Disease Outbreak News. Region: DRC and Uganda. Access: open. Direct source
World Health Organization. “Multi-country outbreak of mpox, External situation report number 68.” Published 31 July 2026. Type: official situation report. Region: global, with Africa focus. Access: open. Direct source
BBC Verify. “Google withdraws new Earth AI tool after warnings over misinformation risks.” Published 31 July 2026. Type: verification reporting. Region: global. Access: open. Direct source
BBC News. “China's new challenge as natural disasters strike - fake AI videos.” Published 27 July 2026. Type: independent reporting. Region: China and East Asia. Access: open. Direct source
Harvard Kennedy School Misinformation Review. “Experts disagree on how to fight AI disinformation, but agree that health and politics need different solutions.” Published 29 July 2026. Type: exploratory expert survey. Region: international, with a Western sample bias. Access: open. Direct source
Harvard Kennedy School Misinformation Review. “Professional and community-based fact-checking show different strengths, but neither performs strongly across trust, scalability, and impact.” Published 30 July 2026. Type: systematic review. Region: international, with platform and study-distribution limits. Access: open. Direct source
Reuters. “Japan quake kills at least two after blast at Kumamoto mall.” Published 29 July 2026. Type: independent reporting. Region: Japan. Access: open. Direct source
NPR. “Death toll from southwestern Japan quake climbs to 34.” Published 31 July 2026. Type: independent reporting. Region: Japan. Access: open. Direct source
Federal Bureau of Investigation and US Environmental Protection Agency. “Malicious Cyber Actors Targeting Water and Wastewater Sector Internet-Facing Programmable Logic Controllers, Causing Operational Disruptions.” Published 30 July 2026. Type: joint public warning. Region: United States. Access: open. Direct source
Associated Press. “2 killed when firefighting helicopters collide near Athens as Europe's wildfire emergency rages.” Published 2 August 2026. Type: independent reporting. Region: Greece and Europe. Access: open. Direct source
CNN. “What we know about the plane crash that killed 11 tourists near Nazca Lines.” Published 2 August 2026. Type: secondary reporting of official and company statements. Region: Peru and Latin America. Access: open. Direct source