# AI Security Failures Threaten Global Digital Supply Chains New research reveals widespread cybersecurity vulnerabilities among AI suppliers impacting enterprise ecosystems. - **Source:** [Latest news](https://brifing.si/source/77d15218-219f-472d-bf40-31b5646d2942) - **Published:** 2026-11-10 - **Topic:** [technology](https://brifing.si/c/technology) - **Original article:** - **On Brifing:** https://brifing.si/s/bee06e7d-4173-4c8d-a6ed-cd987a96864a ## Widespread AI Security Weaknesses A comprehensive 2026 report by Black Kite identifies significant cybersecurity failures across the global AI supply chain. Researchers examined 18,160 companies, finding that many AI suppliers struggle with fundamental security practices. These failures include unpatched critical vulnerabilities, poor credential protection, and inadequate user authentication, which collectively threaten the integrity of enterprise digital ecosystems. ## Network and Botnet Risks The study highlights alarming network-level threats within the AI industry. Researchers detected unusual network activity at 44.3% of the assessed organizations, while 41.8% showed evidence of botnet-related activity. These findings suggest that many AI service providers lack the robust infrastructure necessary to defend against common automated cyber threats targeting their operational environments. ## Enterprise AI Spending Surge Enterprises are rapidly adopting AI, with spending reaching $37 billion in 2025. Approximately 76% of these solutions are purchased from external vendors rather than developed internally. This reliance on third-party providers creates a massive attack surface, as every integrated AI component introduces potential security gaps directly into the corporate infrastructure of the purchasing firm. ## Model Context Protocol Vulnerabilities Analysis of Model Context Protocol components, which link AI applications to business systems, reveals severe security risks. Out of 1,732 logged findings, 48.8% were rated as High or Critical. These vulnerabilities include dangerous code and command injection flaws, which could allow malicious actors to compromise the connection between AI tools and core business platforms. ## Credential Theft and Exposure AI suppliers often hold privileged access to sensitive customer data and developer environments. The research found that 19.9% of AI companies offering enterprise deployment solutions had corporate login information appearing in infostealer records. Overall, 6.8% of all studied AI suppliers were linked to stolen corporate credentials, posing a direct threat to their clients. ## Visibility and Monitoring Gaps Maintaining visibility into the digital supply chain is essential, yet 33.2% of assessed AI companies maintain a limited digital footprint. This lack of transparency regarding domains, subdomains, and IP addresses makes it difficult for businesses to monitor their vendors effectively. Without clear visibility, enterprises cannot accurately assess the security posture of their AI partners. ## Continuous Assessment Strategies Single point-in-time security assessments are no longer sufficient for managing modern AI risks. Experts recommend frequent, ongoing monitoring and active engagement with suppliers to understand their specific technologies and dependencies. Businesses must prioritize managing ransomware-related exposures and vulnerabilities that are already being actively exploited in the wild to mitigate immediate organizational threats. ## Risk Management and Mitigation Enterprises should audit every AI tool’s integrations, access levels, and data reach before deployment. By weighing the potential risk against the reward, companies can decide if a tool is worth the exposure. If kept, organizations should strictly limit the AI tool's permissions and access to only the specific data required for its function. ## Sources covering this story - https://www.zdnet.com/tech/ai-security-failures-supply-chain-risks-black-kite-research/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._