TLDR: Belitsoft, a leader in custom software development and AI security, has released a new report detailing the pervasive challenges and significant risks associated with ‘Shadow AI’ within corporate environments. The report underscores the urgent need for robust AI governance as employees increasingly adopt unauthorized AI tools, leading to potential data breaches and compliance issues.
Belitsoft, a prominent custom software development company known for its expertise in implementing security testing best practices for AI software products, has issued a comprehensive report on ‘The Current State of Shadow AI.’ This timely analysis sheds light on the growing phenomenon of employees utilizing artificial intelligence tools and applications without formal approval or oversight from their organization’s IT departments.
‘Shadow AI’ refers to the unsanctioned deployment and use of AI tools, often by employees seeking to boost productivity or streamline workflows. While seemingly innocuous, this practice introduces substantial risks to enterprise data security, regulatory compliance, and overall corporate reputation. Belitsoft’s report, released in July 2025, aims to provide a critical overview of these challenges.
Industry data from other recent reports, such as one by ManageEngine in early July 2025, corroborates the widespread nature of this issue. This report indicated that nearly two-thirds of decision-makers identify data leakage or exposure as the primary risk associated with shadow AI. Furthermore, more than four out of five tech leaders acknowledge that the rapid adoption of AI tools by employees is outpacing IT departments’ capacity to properly vet these applications for safety. Alarmingly, three in five workers admit to using unsanctioned AI tools more frequently than the previous year, with one-third having entered confidential client data and 37% plugging private company data into external AI systems.
Belitsoft’s broader insights into ‘AI Software Development Trends in 2025’ reinforce the complexities surrounding AI adoption. The company notes the absence of a single global rulebook for AI, leaving organizations vulnerable to existing product-liability, consumer-protection, antitrust, data-protection, and cyber-crime laws if harmful or negligent AI systems are deployed. Belitsoft emphasizes that secure and traceable AI pipelines are mandatory, and highlights unresolved gaps such as the immaturity of tooling for AI explainability and provenance, as well as the fuzzy nature of training-data rights and compensation, which could invite future lawsuits.
Integrating safety guardrails, revamping data governance, and comprehensive staff training are crucial steps that often take longer than initial proof-of-concept demonstrations, according to Belitsoft. This suggests that while the allure of AI-driven productivity is strong, the foundational work for secure and compliant deployment is often overlooked in the rush.
Also Read:
- AI’s Evolution: Shifting Focus from User Prompts to Comprehensive Governance Frameworks
- Algorithmia Unveils Advanced Governance Tools to Tackle Machine Learning Compliance Challenges
The report from Belitsoft implicitly calls for a more strategic and governed approach to AI integration within enterprises. As Ramprakash Ramamoorthy, director of AI research at ManageEngine, stated in a related context, ‘Shadow AI represents both the greatest governance risk and the biggest strategic opportunity in the enterprise. Organizations that will thrive are those that address the security threats and reframe shadow AI as a strategic indicator of genuine business needs.’ Belitsoft’s findings align with this perspective, advocating for organizations to develop robust AI strategies that incorporate strong governance and security initiatives to manage risks effectively while still harnessing the benefits of AI technologies.
