CIOs list securing AI as top risk

Chief information officers are now ranking artificial intelligence security with malware, ransomware and phishing as a top cyber risk, even as the Central Intelligence Agency expands artificial intelligence tools for analysts. (prnewswire.com) (nextgov.com) Logicalis said April 13 that 28% of chief information officers now cite artificial intelligence itself as a major cyber risk, just behind malware and ransomware at 33% and phishing at 30%. The company said 77% of organizations reported a cybersecurity incident in the past year. (prnewswire.com) The same survey said 57% of respondents believe employees put data at risk through artificial intelligence use, and 34% said artificial intelligence has created new security blind spots. Logicalis linked those blind spots to “shadow artificial intelligence,” or staff using tools outside approved systems and audits. (prnewswire.com) Artificial intelligence security means controlling what models can see, what they can generate, and where prompts and data go after an employee uses them. The Intelligence Community’s Directive 505 requires governance for artificial intelligence developed, acquired or used across the community, including risk management and oversight. (dni.gov) That governance question is moving from corporate information technology departments into national security agencies. Central Intelligence Agency Deputy Director Michael Ellis said on April 9 that the agency ran a few hundred artificial intelligence projects last year and recently used artificial intelligence to generate an intelligence report for the first time. (nextgov.com) Ellis said the agency expects “within the next couple of years” to build artificial intelligence coworkers into all of its analytic platforms, with humans retaining control over judgments and decisions. Storyboard18, citing The Wall Street Journal, reported the tools are meant to help analysts draft reports, test hypotheses and spot patterns faster. (nextgov.com) (storyboard18.com) The Central Intelligence Agency has been building this capacity for years through its Directorate of Digital Innovation, which the agency says combines information technology, data, artificial intelligence, cyber defense and open-source intelligence. In a 2023 Studies in Intelligence essay, agency authors said artificial intelligence can help analysts handle more data but warned that opaque systems can complicate transparency and trust. (cia.gov 1) (cia.gov 2) Corporate technology chiefs and intelligence officials are landing on the same operating rule: use artificial intelligence more, but fence it in. The tools are spreading into everyday work faster than the systems for logging, reviewing and limiting them. (prnewswire.com) (dni.gov)