A recent study by SANS Institute and Cylance shows that many cybersecurity professionals believe AI is not yet mature and safe enough to be implemented for security. Survey respondents identified the following 7 risks of such implementation:
- Loss of privacy as a result of AI consuming data
- Over-reliance on one central AI algorithm
- Lack of understanding of algorithm limitations
- Shortcomings in terms of (meta)data protection
- Inadequate training solutions
- Insufficient visibility into AI-based decisions
- Use of incorrect algorithms for problem solving
When asked about the main obstacles to AI implementation, to top issues mentioned were the immaturity of AI (35%); insufficient resources (27%); a lack of funds and management commitment (24%); and perceived risk (10%), while 2% of respondents dismissed AI as marketing hype.
Read more: Top 5 barriers to AI security adoption
