Most of us have gotten into an argument sometime in our lives, based on bad information.
Once we find out the actual truth of a situation (that would have prevented the argument in the first place) it's too late: feelings are hurt, and damage is done.
A new research paper is painting a similar picture when it comes to hacking nuclear arsenals around the world: compromised systems could feed human decision makers false information and potentially lead to a nuclear launch that would not happen otherwise.
By the time the compromise is discovered, it will be too late. Perhaps a new war will have started.
Cyber attack risk is growing for nuclear arsenals around the worldPart of this risk is because most nuclear arsenals around the world were developed decades before the public internet and IoT were even things. And the risk of hacks and attacks is growing, as a result.
"While key risk areas have existed for a long time, new technology has exacerbated these risks. With each new digital component embedded in the nuclear weapons enterprise, new threat vectors may emerge," according to paper authors Dr. Beyza Unal, Nuclear Weapons Policy, International Security Department and Dr. Patricia Lewis, Research Director, International Security Department at the Chatham House think tank.
The new research paper is quite detailed and a very disturbing read. For brevity, here are five key ways a compromised nuclear system could literally "blow up" in our faces or someone else's face.
A hacked system could hurt the ability to:
"Communications as well as the transfer and storage of data are key targets for cyber attackers," the authors say.
They note that currently, nation-states are the most likely to have the capability to hack nuclear launch systems or support networks.
However, there are signs terrorists and organized crime are getting in on the radioactive power play:
"Similar cooperation between organized crime groups and terrorist organizations has already been seen in other areas, such as the illicit trafficking of radioactive and nuclear materials and terrorism related activities, in Moldova and in Georgia. There is an emerging trend that these groups may share an interest in exploiting vulnerabilities in critical national infrastructure and strategic military assets."
Would AI limit nuclear risk and cut down on false alarms in a digital age like we saw at the start of 2018?
Maybe. But only until a system is compromised. Then, automation could deliver disaster.
"These developments will enable machines to make decisions based on learning from experience rather than on pre-programmed responses. Over-reliance on an autonomous system may result in automation bias, where the data is believed without being questioned by both human operators and decision-making machines. Rigorous monitoring of automated information could reduce automation bias but it requires critical and skeptical minds and algorithms that can critically analyze the received information."
Back to the thought at the start of this story: arguments based on bad information can usually be apologized for and damage repaired.
But it would be impossible to undo lives lost from a nuclear missile launched off of bad intel from a hacked system.