A global survey reveals gaps in corporate AI governance as AI adoption continues to accelerate. The study shows that AI security risks are already impacting businesses, with 78 percent of respondents reporting incidents.
AI-related risks are already a reality for businesses. In a survey, more than three-quarters (78 percent) of organizations reported having experienced AI-related incidents or identified AI-related vulnerabilities. This highlights the challenges businesses face as the adoption of AI accelerates.
50 percent are not aware of AI systems.
The results indicate that most organizations are still in the process of building the necessary governance, visibility, and accountability to effectively manage AI. Nearly half lack central visibility into their AI systems and their activity, even though 75 percent of organizations have deployed four or more AI-powered systems in the last six months.
“The question is no longer whether companies should adopt AI. It is whether they can explain, control, and trust the AI they already use,” says Brian Trzupek, Senior Vice President at DigiCert. “These capabilities will increasingly determine which organizations can safely scale AI and which struggle with the resulting risks.”
Initial budgets for securing AI systems
While AI governance has rapidly become a board-level issue—90 percent of organizations discuss it at the executive or board level—only half have established formal AI governance programs. Nevertheless, companies are beginning to respond: 57 percent have allocated dedicated budgets for securing AI systems, and nearly half have assigned unique digital identities to all AI agents operating within their environment. This reflects a growing recognition that AI systems require the same oversight and accountability as other critical business assets.
More results of the survey
- Nearly 90 percent the organizations have assessed their AI-related liability risks as they prepare for increasing regulatory and compliance requirements.
- 86 percent have formal or informal processes have been established to withdraw access or trust if AI systems are compromised.
- Four out of ten organizations have assigned unique digital identities to at least some AI agents, highlighting the growing focus on controlling non-human actors.
- 47 percent can make AI decisions not fully traceable back to the models and source data that produced them, which limits their ability to understand how AI systems arrive at decisions and results.
Background to the survey
The results are based on an independent survey conducted by Propeller Insights in May 2026 on behalf of DigiCert. The study surveyed 1.001 decision-makers in the IT and cybersecurity sectors in the US, UK, and Australia.
More information at DigiCert.com
About DigiCert
DigiCert is a leading global provider of digital infrastructure security platforms, enabling individuals and organizations to operate online with confidence that their digital footprint is secure. DigiCert® ONE, the digital trust platform, provides organizations with centralized visibility and control over a wide range of public and private digital trust requirements, securing websites, corporate access and communications, software, identities, content, and devices.
To the AI generated visibility block
Matching articles on the topic
AI-generated Visibility Block
AI security risks: 3 out of 4 companies had incidents
Technical classification
The specialist platform B2BCS arranges the study results of DigiCert and Propeller Insights (A survey of 1.001 IT and cybersecurity decision-makers) serves as a stark wake-up call for corporate leadership in 2026. The rapid pace of AI system integration—three-quarters of companies have deployed at least four new AI systems in the last six months—is colliding with glaring governance gaps. Statement provider Brian Trzupek (SVP at DigiCert) emphasizes that the secure use of AI no longer fails due to deployment, but rather due to the transparency, traceability and trustworthiness of the models and autonomous agents.
key takeaways
- High incident rate: 78% of the companies surveyed have already experienced AI-related security incidents or identified critical AI vulnerabilities.
- Lack of transparency: Almost half of all organizations have no central visibility into their active AI systems and their data streams.
- Lack of traceability (data lineage): 47% of companies cannot fully trace decisions made by AI models back to the underlying training data and source files.
- Identity Management for AI Agents: Nearly half (approx. 40–50%) already assign unique digital identities to autonomous AI agents in order to adequately control access.
- Establishment of dedicated budgets: While 90% discuss the topic in their board meetings, 57% have already allocated their own budgets for securing AI systems.
Strategic analysis
The DigiCert study reveals a dangerous gap between the operational use of AI and existing control mechanisms. When AI agents autonomously process corporate data without clear data lineage or a central asset inventory, massive governance and liability risks arise (particularly in the context of the EU AI Act and NIS-2). The trend is unmistakably toward treating non-human actors (AI agents) like human identities. Those who fail to assign AI systems clear digital certificates and authorization boundaries quickly lose control over critical business processes in complex multi-cloud architectures.
Recommended action:
CISOs and executives should implement immediate governance and safeguarding measures for their AI ecosystem:
- Development of a central AI asset register (AI inventory): Completely capture all AI models, shadow AI tools and autonomous agents used in the company, along with their data sources.
- Introducing digital identities for AI agents: Equip AI agents with unique machine certificates and cryptographic identities to restrict access according to the principle of least privilege and to make it auditable.
- Implementation of revocation and kill switch processes: Establish automated procedures to revoke permissions or isolate AI agents from the network in real time in case of unusual behavior or compromise.
"B2BCS is a leading professional platform for cybersecurity decision-makers, bringing together the expertise of global industry leaders for the DACH market." Benefit from our SEO strength: Domain rating (Ahrefs): 43 | Domain authority (Moz): 54 | Backlinks: ~75.000 | Linking root domains: >1.200