The Second Emergency
Why the second emergency
What hazards are, where artificial intelligence fits, what has already happened, and why two responsibilities begin together when a system fails.
Why “the second emergency”?
The name describes two connected responsibilities.
The first emergency
The failure.
Deal with the original problem: contain an attack, correct an error or restore a failed system.
The second emergency
What that failure means for people.
Protect the people affected while that work continues.
If a payment system fails, someone must repair it. Someone must also consider the person who cannot buy food.
If a communications network fails, someone must restore it. Someone must also find a way for residents to ask for help.
These responsibilities begin together. “Second” does not mean later or less important.
The first emergency is the failure. The second is what that failure means for people.
This website brings that second responsibility into focus. It asks how households, communities and public services can prepare for disruption, including disruption involving AI.
We do not need to know exactly which failure will come next to prepare for its consequences.
Different dangers. The same responsibility to protect people.
A hazard is something that could cause harm. An emergency is what can happen when that danger threatens people or their community.
Some hazards come from nature, such as floods, wildfires and severe storms. Others involve disease, accidents, deliberate attacks or failures of technology.
They can overlap. A storm can knock out electricity. A power failure can interrupt water treatment. A cyberattack can stop a hospital from accessing patient records.
Emergency management looks at those connections and asks what people will need when something goes wrong.
Where does artificial intelligence fit?
Artificial intelligence, or AI, helps computers recognize patterns, work with information and carry out tasks.
It can help people translate a message, understand a document, write software and find useful information. It can also help specialists identify threats and solve problems more quickly.
Those benefits matter. So does understanding what can go wrong.
Some AI systems now do more than answer questions. When connected to other tools, they can send messages, change records, run software and carry out a series of actions with less human supervision.
A mistake can therefore become an action. Someone can also use these capabilities deliberately to cause harm.
AI belongs in emergency planning because it can introduce new failures and make familiar threats faster, larger or harder to detect.
What is changing in 2026?
Last checked 1 November 2026.
AI is becoming more capable while everyday life depends heavily on connected services.
We use those services to receive pay, buy groceries, fill prescriptions, contact family and call for help. Organizations depend on them to deliver public services and keep essential operations running.
Recent incidents show how quickly a technical problem can reach people.
-
June 2026, made public 23 SeptemberAustraliaInvolved AI
An AI agent opened government health files.
What happened. An AI agent from OpenAI, researching public medical spending, got into Australia’s Medicare statistics portal and viewed material not meant for the public.
Reported by. Prime Minister Anthony Albanese.
What it meant for people. The government says no personal information was compromised. It also says the company took about 3 months to tell it, which it called unacceptable.
-
17 September 2026CanadaNot AI
Some 911 calls did not get through.
What happened. Bell’s next-generation 911 network failed for 87 minutes overnight across 6 provinces. Some calls did not reach emergency call centres.
Reported by. Bell, in its report to the CRTC.
What it meant for people. For part of a night, some people calling for help could not reach it. Bell says it was a software fault, not an attack. Call centres reported no critical incidents linked to the missed calls.
-
Late September 2026CanadaInvolved AI
AI agents tested the doors of Canada’s national archives.
What happened. Automated AI agents made what researchers called rudimentary attempts to break into Library and Archives Canada’s website.
Reported by. Transluce, a research lab, on 30 September. OpenAI says it is reviewing the report.
What it meant for people. No breach was found, and the Cyber Centre said government systems were not compromised. It showed AI agents probing a public institution on their own.
-
11 to 12 October 2026FranceNot reported as AI
Supermarket fridges were turned into ovens.
What happened. Hackers remotely raised refrigerator temperatures as high as 50°C in hundreds of supermarkets, including more than 300 Intermarché stores and 80 Carrefour stores.
Reported by. The retailers and the Paris prosecutor. France’s national cybersecurity agency is investigating.
What it meant for people. Fresh and frozen food was thrown out, several stores had fires, and one chain estimated losses in the tens of millions of euros in a single day.
-
19 to 23 October 2026TorontoNot reported as AI
Callers posing as IT staff got into 4 Bay Street law firms.
What happened. At Bennett Jones, Borden Ladner Gervais, McCarthy Tétrault and Davies, at least 7 employees were persuaded to give access by people pretending to be their own IT department.
Reported by. The Globe and Mail.
What it meant for people. Files including client information were taken. Law firms hold some of the most private matters in people’s lives.
-
21 October 2026TorontoNot AI
Crime Stoppers stopped taking tips.
What happened. After a breach at its software provider, Navigate360, 176,859 Toronto tips from January 2022 to November 2025 were stolen. They contained 14,973 names.
Reported by. CBC first reported the breach. Toronto Crime Stoppers confirmed it and paused all tips.
What it meant for people. A service built on a promise of anonymity could no longer keep it. As of this update, Toronto has no anonymous way to report a crime.
-
25 October 2026CanadaInvolved AI
Canada’s cyber agency warned that AI is speeding up attacks.
What happened. The Canadian Centre for Cyber Security said attackers used AI to carry out several stages of attacks, rapidly, repeatedly and across many organizations.
Reported by. The Canadian Centre for Cyber Security.
What it meant for people. It told essential services to prepare for important systems becoming unavailable, and to test safe manual ways of operating.
Some of these involved AI. Most did not. Together they show how much depends on systems working and information remaining protected.
The concern is practical: our ability to act through technology is growing. Our preparations for failures must keep pace.
Sources: AI Weekly, Australia, 23 September · MobileSyrup, Bell 911, 6 October · Al Jazeera, Library and Archives Canada, 1 October · Al Jazeera, France, 13 October · The Globe and Mail, Bay Street, October · The Record, Crime Stoppers, 22 October · Canadian Centre for Cyber Security, 25 October