Logiciel Solutions Contact Us
Success Stories Tech News Investors Contact Us
framework

AI Security Vendor Scorecard.

Every AI vendor's security page says the right words. A questionnaire cannot distinguish a vendor who built these controls from one who knows what they are called, which is why most vendor security reviews end up measuring how good the vendor's security writer is. This scorecard puts five knockouts first, weights seven dimensions by real exposure rather than ease of assessment, and then hands you an architecture interview with the vendor's engineer, because that is the part of the process that actually separates built from described.

In depth

Multi-Tenant But Fully Isolated, With No Layer Named.

01

What happens by default: the questionnaire names every control correctly.

The vendor says multi-tenant but fully isolated, without naming the layer where tenant identity is enforced. They say the model is trained to ignore injected instructions. They say the agent only does what you ask it to, which describes intent rather than permissions. Every answer is defensible and none is checkable, and the assessment ends up measuring how good their security writer is.

In shortEvery answer is defensible and none is checkable, an…
02

What good assessment does: it asks to speak to an engineer.

Somebody who works on the system, not a sales engineer. If the vendor will not provide one, that refusal is scored as a finding rather than worked around. Then it listens for specificity, because a person who built the thing answers with architecture, constraints and trade-offs while a person who read the security page answers with control names. Ask what happens to a prompt from your tenant from API to storage and where tenant identity gets enforced: a strong answer names a layer below the application and mentions caches and logs unprompted.

In shortAsk what happens to a prompt from your tenant from A…
The detail

Where Vendor Security Reviews Usually Miss.

Three gaps recur across AI vendor assessments, and all three are invisible to a standard security questionnaire.

Zone · 01

Four leak paths

not one

Vendors answer isolation questions for the primary database and assume the rest is covered. AI systems leak between tenants in four other places: prompt and completion stores, vector indexes, response and context caches keyed in ways that can collide, and log stores a support engineer can read. A vendor who has not considered all four has not designed isolation.

Zone · 02

Scope by deployment model first

For a hosted model provider you assess their infrastructure and isolation, because you own everything above the API. For AI-native software you assess the full stack, because they built the prompt assembly and tool layer you would otherwise own. For AI bolted onto an existing product, ask when the AI feature was security reviewed and by whom.

Zone · 03

The refusal is the signal

A vendor who will not share independent testing evidence, even under a non-disclosure agreement, is telling you something precise: either they have not been tested, or they have and would prefer you not read it. How routinely they handle that one request predicts the rest of the assessment better than any individual answer.

By the numbers

The figures that make it a board-level conversation.

45 min
architecture interview, the part that separates built from described
20
the weight on tenant isolation, because that failure triggers your notification duty
1 in 4
malicious breaches are now AI-enabled, at an average cost of $6M
Inside the report

What you'll take away.

01

Step 1 - Scope by deployment model before you start

A hosted model provider, AI-native software, AI bolted onto an existing product and an agent platform need different questions and different weightings.

02

Step 2 - Run the five knockouts in one call

Tenant isolation across all four leak paths, encryption of prompt data, single sign-on with per-user access, testing evidence under NDA, and an agent permission model.

03

Step 3 - Book the architecture interview with an engineer

Ten questions, each paired with what a strong answer sounds like and what a weak one sounds like. Not a sales engineer, and record the refusal if you cannot get one.

04

Step 4 - Score, then apply the three override rules

Weak tenant isolation is a decline. Weak application security on an agent platform is a decline. No adversarial testing lowers confidence in every other dimension.

Questions

Frequently asked.

What should we ask an AI vendor about security specifically?

Start with tenant isolation across all four AI leak paths: prompts, embeddings, caches and logs, not just the database. Then whether prompt and completion data is encrypted in transit and at rest, whether they offer single sign-on with per-user access control, whether they will share independent testing evidence under NDA, and what the permission scoping model is for any agent features.

How do we assess a vendor we cannot audit?

Substitute the architecture interview. Ask for an engineer rather than a sales engineer and listen for specificity: a strong answer names a layer, a constraint or a trade-off, while a weak one names a control. If they will not provide an engineer, score the application security dimension from the questionnaire alone and record the refusal as a finding.

What is the single most useful question to ask?

What happens when your guardrail service is unavailable. It is narrow, factual, and cannot be answered from a security page. A vendor who has genuinely engineered their AI security has decided it and will say so in one sentence. One who assembled a stack will say they would have to check, and the answer is almost always fail-open.

Does a SOC 2 report cover AI-specific risks?

Not the ones that matter here. It is meaningful evidence for infrastructure, access control and incident response, and you should ask for the full report rather than a bridge letter. It will not tell you whether retrieval enforces authorisation at query time, whether vector indexes are partitioned by tenant, or whether agent tool scopes are bounded.

How should we score a vendor who has never been red teamed?

Score that dimension no higher than one, and note explicitly that it lowers confidence in every other dimension rather than just its own, because nothing has been independently verified. Adversarial testing is the best available proxy for whether the other six dimensions are real or aspirational.

Is this needed for software that just added an AI feature?

Especially then. That category has the most common gap: AI bolted onto a product whose security model predates it, often without the AI feature going through the same review as the rest. Ask directly when the AI feature was security reviewed and by whom, and weight tenant isolation, application security and adversarial testing heavily.

Get the framework

Have it emailed to you.

Drop your details and we'll send AI Security Vendor Scorecard straight to your inbox - no spam, unsubscribe anytime.

Download framework
Next step

Ask what happens when their guardrail is unavailable.

It cannot be answered from a security page. Bring us a vendor you are evaluating and we will sit in on the architecture interview. A working session, not a sales pitch. SECTION 7 - FAQ - 5 to 8 questions

Talk to our engineers