AI違規致損$5.33M — 92%的企業卻缺少基礎訪問控制

ibm's latest research with the ponemon institute across 602 companies found that 92% of organizations experiencing ai security breaches failed to implement adequate access controls. incidents involving ai cost an average of $5.33 million, significantly higher than the global average of $4.99 million, and peaked at $6.04 million when attackers themselves utilized ai tools. the study highlights that security failures typically stem from compromised apis, connected applications, or misconfigured cloud services rather than the ai models themselves, regardless of whether they are open-source or proprietary. this suggests that enterprises are prioritizing model adoption over foundational security protocols, creating expensive vulnerabilities.