The study highlights growing concerns about AI-generated attacks, with 54% identifying them as their biggest security risk