Cybercrime and AI: threats and opportunities

The most common types of attacks include phishing: attackers break into the email account of a high-level executive and then sell sensitive data on the dark web, possibly blackmailing the company. Ransomware is also a serious problem, for instance a ransomware attack on the British food delivery company Compass Group – they employ more than 500,000 people – has crippled the entire supply chain. Distributed denial- of-service (DDoS) attacks are also a problem. In these hackers generate so much traffic to the target server that it becomes unable to function normally. The online platforms of FMCG companies such as webshops and customer relation systems can be particularly vulnerable, as they can interrupt continuous production and logistics. Data poisoning is when AI systems are deliberately trained with fake or manipulated data, which can cause big problems later on, as the poisoned AI systems may react to user interactions in unforeseen ways.

Prevention and crisis management

In cybercrime situations there is great panic and suddenly it isn’t enough to be a security expert – one also needs to become a communication specialist. It isn’t good if one says nothing to the outside world, but it is also bad to cause alarm, which can scare off buyers and hurt share prices. Companies need to use up-to-date security software and strong data security protocols: data encryption, regular backups, and strict compliance with data protection policies… It is a good idea to educate staff regularly, to make them aware of the importance of multi-factor authentication and it is key to have a crisis management plan in place. The spread of AI isn’t limited to traditional “machine” areas; it is also playing an ever-bigger role in the creative industries.

Neither a tool, nor a human

Arthur Keleti’s presentation also discussed how AI is influencing human interactions, especially in areas where we can use AI instead of humans. For example in a healthcare simulation AI has been able to make diagnoses and recommend treatments, suggesting that in the future AI may be able to support or even partially take over certain areas of the work of healthcare professionals. The expert spoke about to what extent AI can display “human” qualities such as empathy, creativity or integrity. It is clear that AI isn’t simply a tool, but it is clearly no human either, which raises new challenges in terms of ethical judgement and legal regulations. //