A survey of large enterprises around the world suggests mixed awareness of AI-related security vulnerabilities amid rising data complexity, operational challenges.

Based on a survey of 650 global^ IT decision-makers* (including additional research with 150 AI strategists from organizations with at least US$200m in annual revenue) on enterprise AI readiness and application strategy, a firm specializing in application delivery and security has released some findings on the state of AI adoption and associated operational, security, and governance practices to the media.

Second, 70% of the respondents marked with moderate readiness had indicated active use of generative AI, with the remainder reporting their organizations as working toward deployment. Respondents characterized AI use in 25% of enterprise applications overall, with higher saturation in portfolios deemed highly ready and lower, more siloed or experimental use among those classed as low-readiness.

Other findings

Third, 65% of all respondents had indicated they were using two or more paid AI models along with at least one open-source model, with the average reported use of three models. Paid models such as GPT-4 were most commonly cited for active use, with open-source models including Meta’s Llama variants, Mistral AI variants, and Google’s Gemma also reported. Also:

• 71% of respondents indicated use of AI to augment security practices at the time of the survey.
• 18% of survey respondents categorized as moderately ready cited the deployment of an AI firewall, with an additional 47% reporting intentions to deploy such measures within twelve months.
• 24% of all survey respondents reported conducting continuous data labeling for governance purposes.
• 79% reported repatriation of at least one app from a public cloud to on-premises or colocation environments.
• 58% identified API-related complexity as a notable pain point, and 53% cited inconsistent security policies across deployment sites.

According to the firm that commissioned the survey, F5, Inc., the results were positioned as identifying opportunities for enterprises to improve secure, scalable AI adoption. Among the highlighted trends, several persistent obstacles suggest that projections of seamless AI integration and readiness should be approached with caution. As with any industry-sponsored report, readers are encouraged to weigh the findings critically and consider the broader context before making decisions based solely on the conclusions presented.

*with 65% comprising executive or senior leadership roles (executive IT leadership, senior non-IT leadership, platform/operations, network/NetOps, security/SecOps, cloud, data science/machine learning, architect, and others). All data sets were self-reported by survey respondents, not independently verified, and reflect the beliefs, reporting, and practices of respondents at the time of the survey. The survey did not provide details on geographical coverage^, sampling method, sampling frame, response rates, survey design, question wording, or weighting adjustment