Cybersecurity slows down the introduction of AI agents
Companies around the world have high expectations for AI agents. However, not all of them are ready to implement them. According to a Salesforce study, many still need to resolve issues related to cybersecurity and compliance.
AI agents are on the rise - and they will have a big impact. At least that's the view of many of the managers surveyed on behalf of Salesforce. Cybersecurity is no exception, as Salesforce has now revealed in a new study. All of the 2,000 managers and IT security officers surveyed worldwide see at least one security-relevant area of application that could be improved with agent-based AI. However, the study also shows that many companies still have some homework to do in the area of cybersecurity before introducing the AI wonder weapon.
7 out of 10 Swiss companies increase their cybersecurity budget
According to the study entitled "State of IT Report", 74 percent believe that their security practices are in need of a fundamental transformation. 48 percent state that their data infrastructure is not sufficiently prepared for the use of agent-based AI. More than half (55 percent) express doubts about having the necessary guard rails and security mechanisms in place. In addition to known threats such as phishing, malware and cloud risks, IT managers are increasingly concerned about data poisoning - i.e. the targeted manipulation of training data by third parties, Salesforce summarizes.
With 50 people surveyed in this country, the study's database for Switzerland is rather thin. However, it indicates that the majority of companies here are reacting to the increasing risk and increasing their cybersecurity budget. 70 percent of the managers surveyed in Switzerland stated this. In addition, 30 percent of IT security teams in Switzerland already use AI agents in their day-to-day business - and the trend is rising, Salesforce adds. According to the survey, more than twice as many expect to implement them within the next two years. Among other things, they expect this to result in faster threat detection, improved audits of AI models and greater automation of operational processes.
Compliance as an obstacle
4 out of 5 IT security managers surveyed in Switzerland see AI agents as a means of improving compliance with data protection laws. At the same time, many respondents (72%) cite regulatory complexity as an obstacle to the introduction of AI agents. Only 52% are confident that they can use AI agents in compliance with regulations today and 66% have not yet fully automated their compliance processes.
In the press release accompanying the study, Salesforce calls for robust data structures and clear governance guidelines. These are essential for AI agents to develop their full potential. (René Jaun/fsi)
