At the annual IoT Security Foundation conference in London, UK, Arm's Chief Architect Richard Grisentwaite recalled one of the key IoT security challenges in his opening keynote: "70% of vulnerabilities (CVEs) are memory-related."
This will become an even bigger problem in the future as AI will show attackers where these vulnerabilities are, putting the system at risk.
John Moore, managing director and co-founder of the IoT Security Foundation (IoTSF), said that memory security needs to be addressed.
"We need to utilize tools such as the Memory Security Error Correction (MTE) feature," Grisentwaite said.
In a presentation at the IoTSF conference, Anna Knack, a research fellow at the Alan Turing Institute's Center for Emerging Technologies and Security (CETaS), outlined research in this area.
Despite the wide range of methods and technologies related to autonomous cyber defense, the researchers settled on the method of machine learning (RL). This decision was made by the good results that have been achieved with RL in other problem areas.
In her presentation, Knack emphasized that the research is still in its early stages and is still many years away. She said that "there are currently no experts in cybernetic artificial intelligence.