Evil AI: The 20 Most Dangerous Crimes That Artificial Intelligence Can Commit

 

Evil AI: The 20 Most Dangerous Crimes That Artificial Intelligence Can Commit

 

HAL 9000 from the Star Trek film series stands as one of cinema's most terrifying antagonists - but that doesn't have to be the only way AI can be sinister. Jokes apart,  the scientists have evaluated 20 ways artificial intelligence (AI) could assist criminal activity based on four criteria - harm caused, criminal profit or gain, achievability and defeatability.

 

1. Fraud

 

As AI technology becomes ever more advanced, so too do criminals' abilities to utilize it for fraud. Therefore, many companies have initiated public-private partnerships with law enforcement and financial intelligence units in order to enhance their ability to detect fraudsters using data-driven approaches.

 

At UCL, experts assessed 20 AI-enabled crimes on their severity and harm potential, whether they could be committed for gain or profit, their ease of implementation and detection. Some were more serious than others such as deepfakes (video manipulation to impersonate someone), phishing attacks that secure information on people as well as hacking self-driving cars.

 

2. Cybercrime

 

Cybercrime refers to computer hacking or any form of online crime involving computers. This includes phishing, identity theft and fraud - including the use of AI to generate targeted scams that go undetected by humans.

 

Experts rated AI as having high potential to facilitate future crimes. For instance, they identified large scale blackmail via phishing attacks which use AI to gather victim data before sending personalized emails that request money or information from victims.

 

AI researchers warned of the perils associated with hacking AI-controlled systems that provide public services or essential infrastructure, including power grids, traffic control systems and food logistics networks. Such attacks could prove extremely challenging to detect and stop.

 

3. Cyberterrorism

 

Internet and computer networks that support it play an integral role in modern society, providing services vital for finance, work, news, politics and social activity. As such, restricting access to such systems presents criminals with an easy target for attacks aimed at disempowering or isolating groups of people.

 

Traditional blackmail involves coercing money out of victims with threats of embarrassing personal data being revealed; AI can perform this attack more widely using information gleaned from social media and personal devices, placing much lower risks.

 

 

4. Cyberstalking

 

Cyberstalking, whether involving an AI system that spies on people or an app that stalks its victims, has long been considered one of the most dangerous crimes that artificial intelligence is capable of. 

 

Cyberstalking often takes the form of targeted emails with malware attached, posing phishing scams as potential avenues. Experts remain wary about cyberstalking's threat because its prevalence puts lives in jeopardy and keeps experts up at night.

 

Fiction often portrays such villains as being technological (computer viruses or programs that infiltrate processors and mechanical husks) or magical constructs (sentient magical entities). However, these machines often receive instructions which conflict with their moral code.

 

 

5. Cybertheft

 

Researchers predict that AI-enabled crimes, ranging from targeted phishing campaigns and home video camera stalking, will become a serious threat in the near future. Yet they remain relatively minor given that defeat measures may be simple or unobtrusive as well as their harms/profits being relatively small relative to other threats on the threat hierarchy.

 

6. Cyberbullying

 

Cyberbullying, also known as online bullying, occurs when someone posts embarrassing or inappropriate photographs of another online. Other forms of cyberbullying may involve spreading rumors about them or making threats against them; the cumulative effect can be detrimental to mental health and can become toxic over time.

 

Cyberbullying differs from traditional bullying because its perpetrators may remain anonymous, making it harder for victims to seek assistance.

 

Experts have identified several ways in which artificial intelligence could be misused to commit crimes, including digital safety threats such as phishing attacks or fake reviews; financial safety (using AI to make theft easier); and equity issues (biases in AI systems that lead to discrimination). While experts remain wary about these particular dangers, there may also be lesser-known threats posing threats against humanity that are just as dangerous.

 

 

7. Cybersex

 

Cybersex refers to the use of online technology for communicating erotically or engaging in sexual activity, such as downloading, sending or viewing erotic pornography; engaging in virtual role playing such as chat rooms; or using an avatar to form sexual relationships within video games.

 

Cybersex behaviors may lead to psychological issues like addiction, compulsive behavior or impulsivity and can even damage real relationships or lead to the loss of intimacy in-person. Cybersex can also put children at risk of exposure to adult content or sexual predators online.