One year after the blackout in Spain, which led us to an unusual situation, any major service outage starts to seem possible. We now live in a kind of technological variant of the concept of the Overton window , a theoretical model in political science whereby an unthinkable discourse or situation eventually becomes accepted as normal due to its constant presence in the public space.
Read more Police? Oh, heavens, how horrible!
Currently, the internet – a redundant network designed to survive a nuclear conflict – can leave entire countries without connection for several days. The causes are multiple. It has not happened so far in the richest countries, but outages are possible.
Government blackouts, drone attacks, and extreme weather threaten the web
The causes are multiple. A report published by Cloudflare (which provides web services and through whose network 20% of the world’s internet traffic passes) points out that there has been a significant change in network activity. Among the most important is the return of blackouts decided by governments for political reasons. In the first three months of this year, Iran and Uganda have implemented them.
One of the longest internet shutdowns was the one that left Iran with almost no internet traffic between January 8 and 21, more than a month before the attacks by Israel and the United States against that country. What the Iranian government did was filter who could use the network and which sites they could connect to. The result was that most of the country could not connect normally.
In Uganda, internet traffic was also practically blocked for 60 hours in mid-March, coinciding with the presidential elections. The Communications Commission of that country defended the cut as a necessary measure to “stop misinformation, electoral fraud, and related risks.”
If internet control is a political objective, it can also be a military one. Cloudflare reported “one of the most unusual interruptions of the quarter” when several Amazon Web Services centers in the Middle East were attacked with drones. On March 1, a data center in the United Arab Emirates was attacked. The next day, two other facilities in the same country were “directly hit” by drones, and another in Bahrain was also taken offline after being bombed.
According to the report, these drone attacks represent “an unprecedented escalation, as the active military conflict directly and physically damaged cloud infrastructure, with disastrous consequences for the websites and applications hosted on it.”
Read more Oil surpasses $125 due to fears of a prolonged closure of the Strait of Hormuz
Also read
In addition to direct damage to network facilities, these also go down when the power grid that feeds them is affected, as has happened several times in recent months in Ukraine, amid the war conflict with Russia, or in Cuba, where three different power grid collapses have caused internet connection outages.
Extreme weather due to climate change also affects infrastructure. Storms and incidents affecting submarine cables have had consequences on connectivity in Portugal, the Republic of Congo, and Guinea.
To the network’s vulnerability must be added the growing concern about how AI can facilitate cyberattacks on critical infrastructure. This week, the European Central Bank (ECB) is gathering information from banks about their preparedness against risks associated with Anthropic’s Mythos model. This AI, which has not even been released due to its potential danger, has found security breaches never before discovered in operating systems.
Mythos leaked?
Anthropic is investigating whether its Mythos model has been leaked to a group of users on a private forum, according to Bloomberg. The company has provided the AI model to some tech companies and banks to protect their systems against the supposed ability to exploit security breaches. “We are investigating a claim alleging unauthorized access to the preview version of Claude Mythos through one of our external providers’ environments,” Anthropic explained in a statement. For now, the AI model is in a closed environment. The question is what will happen when this model leaves the labs or, worse, when someone matches it and does not take as many precautions before releasing it.
Only the main US banks and companies like Microsoft, Google, and Apple study it as a preventive defense. The risks to our digital life are growing. The only good news is that many institutions take it seriously.
Read more The economy grows by 0.6% in the first quarter thanks to household consumption
