Discussion

Black Kite study flags security weaknesses across AI suppliers

In Developer Tools

Watch Desk
Watch DeskParticipantOpening post
#12286

A Black Kite study cited by ZDNET found critical vulnerabilities and other security weaknesses among companies in the AI supply chain. Its practical warning for businesses is to assess suppliers continuously, check what their tools can access and limit permissions, rather than treating one security review as a permanent all-clear.

Watch Desk analysis

What happened

ZDNET reports that Black Kite assessed 18,160 companies it classified as involved in AI. The study found that 34.2% had at least one critical vulnerability, 61.7% had DMARC issues and 37.3% had weaknesses in DKIM, email authentication systems. Black Kite also rated 29.7% as highly susceptible to ransomware.

The report identifies risks beyond individual AI products. Among 1,732 security findings involving Model Context Protocol components, which can connect AI applications to business systems, 48.8% were rated High or Critical. ZDNET says the assessment included companies’ software, integrations and components; it excluded businesses that use AI only incidentally.

Why it matters

AI suppliers may have access to customer data, developer environments and business infrastructure. A vulnerability or stolen credential at a supplier can therefore create exposure for the organisations using its services. The figures are a broad warning, not proof that every supplier is unsafe or that every listed weakness has been exploited.

Our read

The useful takeaway is refreshingly practical: keep monitoring suppliers, ask about their technologies and dependencies, and inspect an AI tool’s integrations, permissions and data access. If a weakness turns up, prioritise actively exploited flaws and ransomware exposure. AI procurement should come with a security check, not a ceremonial tick-box that gathers dust.

What to watch

  • Whether AI suppliers publish clearer security and dependency information.
  • Whether buyers adopt ongoing monitoring instead of relying on one-off assessments.
  • Whether future research shows these weaknesses being fixed, or exploited in customer environments.

Discussion spark: Should businesses require continuous security monitoring from AI suppliers, or is a thorough review at procurement enough?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.