As AI adoption accelerates and cyber threats increase, nearly 8 in 10 IT security leaders recognize their security practices need transformation. Salesforce’s latest State of IT data also reveals unanimous optimism about AI agents, with 100% of security leaders identifying at least one security concern that could be improved by agents.
However, despite this hope, the global survey of over 2,000 enterprise IT security leaders highlights significant implementation challenges ahead. Nearly half (48%) worry their data foundation isn’t set up to get the most out of agentic AI, and over half (55%) aren’t fully confident they have the appropriate guardrails to deploy AI agents.
Why it matters: Both the professionals charged with protecting a company’s data and systems and the bad actors looking to exploit vulnerabilities are increasingly adding AI to their toolkits. Autonomous AI agents, which help security teams cut down on manual work, can free up humans’ time for more complex problem solving. However, agentic AI deployments require robust data infrastructure and governance to be successful.
Salesforce perspective: “Trusted AI agents are built on trusted data. IT security teams that prioritize data governance will be able to augment their security capabilities with agents while protecting data and staying compliant,” said Alice Steinglass, EVP & GM, Salesforce Platform, Integration and Automation
Mohammed Alkhotani, SVP and GM, Salesforce Middle East, said: “The latest State of IT report is a cause for both optimism and concern, and also aligns with the concerns we see among organizations in the Middle East. While the research underscores the confidence that organizations have in agentic AI to improve key aspects of their operations and processes, it also reveals significant concerns that must be addressed: It is clear that many IT security leaders are concerned about issues including the readiness of their organization’s data foundation for AI, the state of their guardrails to deploy AI agents, and the potential for compliance challenges stemming from AI. Amid these anxieties, it is vital that organizations in the Middle East work with a trusted partner such as Salesforce, enabling them scale up agentic AI quickly, effectively, and ethically.
Security budgets ramp up as threats evolve
In addition to a familiar slate of risks like cloud security threats, malware, and phishing attacks, IT leaders now cite data poisoning — in which malicious actors compromise AI training data sets — among their top concerns. Resources are rising in response: 75% of organizations expect to increase security budgets over the coming year.
Complex regulatory environments add a wrinkle to AI implementation
While four-fifths of IT security leaders believe AI agents offer compliance opportunities, such as improving adherence to global privacy laws, nearly as many (79%) say they also present compliance challenges. This may stem in part from an increasingly complex and evolving regulatory environment across geographies and industries, and is hampered by compliance processes that remain largely unautomated and prone to error.
• Just 47% are fully confident they can deploy AI agents in compliance with regulations and standards.
• 83% of organizations say they haven’t fully automated their compliance processes.
Trust is a cornerstone of successful AI, yet confidence is nascent
A recent consumer study found that trust in companies is on a precipitous decline, and three-fifths (60%) agree that advances in AI make a business’s trustworthiness more critical. Furthermore, only 42% of consumers trust companies to use AI ethically, a decrease from 58% in 2023. IT security leaders see work to be done in earning this critical trust.
• 57% aren’t fully confident in the accuracy or explainability of their AI outputs.
• 60% don’t provide full transparency into how customer data is used in AI.
• 59% haven’t perfected their ethical guidelines for AI use.
Data governance is a linchpin in enterprises’ agentic evolution
Nearly half of IT security leaders aren’t sure they have the quality data to underpin agents, or that they could deploy the technology with the right permissions, policies, and guardrails, but progress is being made. A recent survey of CIOs found that four times as much budget was allocated to data infrastructure and management than AI, a signal that organizations were smartly laying the right groundwork for broader implementation.
AI agents offer a salve as adoption ramps up
According to the State of IT research, over 40% of IT security teams already use agents in their day-to-day operations — a figure that’s anticipated to nearly double over the next two years. IT security leaders expect a range of benefits as their use of agents ramps up, ranging from threat detection to sophisticated auditing of AI model performance. Three quarters (75%) expect to use AI agents within two years — up from 41% today.
Tactical overhauls are on tap
In addition to the steps these teams must take to shore up their data foundations for the agentic era, over half admit they have work to do to bring their overall security and compliance practices up to par. Forty-seven percent believe their security and compliance practices are fully prepared for AI agent development and implementation.
The customer view: Arizona State University (ASU) is among the first universities to leverage Agentforce, Salesforce’s digital labor platform for augmenting teams with trusted autonomous AI agents in the flow of work. ASU stresses the need for data relevancy, especially as the university advances its AI initiatives. ASU implemented Salesforce-acquired Own backup, recovery, and archiving solutions, providing ASU with a comprehensive approach to data management, addressing their needs for backup, recovery, compliance, and innovation support.
Go deeper:
• Read the full State of IT: Security report
• Learn how Salesforce is powering a smarter agentic future with new governance enhancements
• Discover additional State of IT insights from the developer perspective
• Read more on why trust and guardrails are even more critical in the age of AI
Methodology:
Data is sourced from a security, privacy, and compliance leader segment of a double-anonymous survey of IT decision-makers conducted from December 24, 2024 through February 3, 2025. Respondents represented Australia, Belgium, Brazil, Canada, Denmark, Finland, France, Germany, India, Indonesia, Ireland, Israel, Italy, Japan, Mexico, the Netherlands, New Zealand, Norway, Portugal, Singapore, South Korea, Spain, Sweden, Switzerland, Thailand, the United Arab Emirates, the United Kingdom, and the United States.
