Experts Release Report on the Dangers of AI
Updated: Feb 24
The rapid development of AI brings with it the potential for new types of security threat to not only humans but other AI’s and computer programs. Researchers from the University of Oxford, Cambridge, Yale, Stanford, OpenAI and the Center for New American Security amongst others have recently published a paper on what they perceive that main types of threat will be and suggests ways in which these threats may be mitigated. As Artificial Intelligence improves the new threats it brings to cybersecurity can be classified into three buckets. The first is the expansion of the existing threats, as the cost of attacks may be lowered by using AI systems to complete tasks which humans would normally do. The second is that there may be new types of threats stemming from AI’s being used to conduct cyber-attacks which humans would be incapable of doing themselves. The last area of threat comes from the ability of an AI to change the character of the threat itself and to make more finely targeted attacks which are more difficult to attribute to any one source. AI’s can be used in this way to make attacks on Digital, Physical and Political security the report goes on with the ability to create fake news and videos.