ZDNET’s key takeaways
- Impressive AI applications can harbor inherent security risks.
- AI companies worldwide are failing to reach basic security standards.
- Supply chain monitoring is paramount when adopting new AI tech.
Artificial intelligence companies worldwide are bringing a new wave of vulnerabilities and security issues to the digital supply chain, researchers warn.
The AI supply chain and security problems
AI and smart solutions don’t automatically equate with safe solutions. In a study titled “2026 AI Supply Chain Report: The Risks Behind Your AI Relationships,” Black Kite says that many AI suppliers — the companies that develop and deploy AI services to today’s businesses — are struggling with “basic security practices.”
Also: Rogue AI incidents hit ‘tens of thousands’: Can businesses trust these tools?
According to the research, after examining the companies themselves, along with their software, integrations, and components, researchers found failures to patch critical vulnerabilities, protect credentials, authenticate users and email, and address network security issues.
The study analyzes 18,160 AI companies worldwide, ranging from 0-10 employees to more than 1,000 employees. A company’s AI involvement was determined through reviews of company products and services and primary business activities, covering six broad groups: everyday application development; AI development and implementation; AI applications in specialized sectors such as healthcare, IT, and finance; industrial organizations; and AI foundations working on data, compute, and modeling.
A small percentage of assessed companies are considered “unclassified,” and the study excluded businesses that use AI only incidentally.
The most important AI supplier risk statistics
- Critical, unpatched vulnerabilities: While critical flaws do not always translate to critical business risk, they should be subject to a risk review, depending on their nature. Still, 34.2%, or one in three, of the companies assessed had at least one critical vulnerability.
- Email security: In total, 61.7% of AI companies had DMARC issues, while 37.3% had DKIM weaknesses, both of which could be abused for impersonation and phishing.
- Ransomware threats: 29.7% of companies were highly susceptible to ransomware. According to the firm’s susceptibility index, almost a third of AI suppliers operate in environments with “heightened exposure” to ransomware attacks.
In addition, unusual network activity was detected at 44.3% of organizations, and botnet-related activity was recorded at 41.8% of AI companies.
Why do AI supplier risks matter?
AI’s main business appeal is its ability to integrate with existing software and platforms to handle manual tasks, heavy lifting, and tedious, time-consuming work, freeing employees to focus on work that requires human oversight.
Enterprises are investing billions of dollars in AI. Menlo’s 2026 research estimates that enterprise AI spend reached $37 billion in 2025, with half of overall spend going to AI apps. In total, 76% of enterprise AI solutions are purchased rather than built internally.
However, the moment an AI or automated component is introduced into a corporate ecosystem or operational platform, any security problems it has impact the user — and each additional integration, such as software components or open-source tooling, may also increase the attack surface.
In an accompanying analysis of AI firms’ use of Model Context Protocol (MCP) components, which can link AI apps to business systems, 48.8% of 1,732 logged security findings were rated High or Critical, including code and command injection vulnerabilities.
Also: The new enterprise AI expert every company needs
It’s not just the AI application itself. As noted by Black Kite, AI suppliers often have access to customer data, developer environments, business infrastructure, and enterprise platforms. If the supplier’s credentials are stolen, this could expose its customers to a range of cyberattacks.
The researchers found, for example, that 19.9% of AI companies offering solutions for enterprise AI deployment had corporate login information in infostealer records, and that stolen corporate logins, overall, were linked to 6.8% of AI suppliers in the study.
Visibility matters, too, for businesses that should be monitoring their digital supply chains. However, 33.2% of AI companies assessed have a “limited” digital footprint, including their domains, subdomains, and IP addresses.
How should businesses respond to AI supply chain risks?
The same principles for protecting the digital supply chain apply to AI suppliers, where continual monitoring and assessment are among the best protections businesses have.
Staying informed and aware of the risks to your own business makes all the difference. As recommended by Black Kite, it takes more than a single point-in-time security assessment — frequent monitoring and engaging with suppliers to understand their technologies and dependencies can reveal potential security issues.
Also: Do you need AI agent liability insurance?
If you find weaknesses, manage them by priority, handling ransomware-related exposures and vulnerabilities actively exploited in the wild first.
Also audit an AI tool’s integrations and components, access, what data it can reach, and the potential impact of exploitation. Once you’re armed with this information, it’s possible to weigh risk and reward — and whether the AI tool is worth the potential attack exposure it brings. If you decide to keep it, limit its permissions and access to what it requires.