CIOPages
All RFP questions

RFP Questions · AI Governance

AI safety & responsible AI questions to ask a software vendor

89 ai safety & responsible ai questions for the RFI, the RFP and the finalists, with scoring guidance, in one Excel workbook.

What this module is for

These are the ai safety & responsible ai questions to put to any software vendor, whatever the product does. Add them to an RFP next to the questions about the product itself.

What it covers. Questions on the vendor's AI safety controls: defenses against prompt injection and unsafe tool use, red-teaming, vulnerability disclosure, harm reporting, abuse mitigation and guidance for safe deployment. Agent permissions are in Agentic safety & autonomy controls.

A selection usually runs in three rounds. The module has questions for each:

  • RFI, to the long list. 22 questions screen out vendors that fall short.
  • RFP, to the shortlist. 35 questions ask for detail.
  • Deep dive, to the finalists. 32 questions ask for evidence.

Each question comes with why it matters, what a good answer looks like and the red flags. The workbook scores the answers 0 to 5, weights them and totals them by round and section.

3 questions from the module

From the RFI round. The first shows part of the guide each question carries; the workbook adds follow-ups, a response format, a weight and a score column.

1. Describe the platform-level controls your product provides to mitigate direct prompt injection (malicious instructions in the user-supplied prompt).

Why it matters. OWASP lists prompt injection first (LLM01:2025) in its Top 10 for LLM Applications 2025. A vendor that pushes the entire defense burden to the customer leaves the buyer to build it. Understanding the platform-side mitigations sets a baseline for the shared-responsibility model.

Good answer
  • Names specific platform mechanisms (e.g., system-prompt isolation, instruction hierarchies, classifier-based detection).
  • Acknowledges that no defense is complete and references known residual risk.
  • Differentiates between platform controls and customer-configurable controls.
Red flags
  • Claims complete or comprehensive prevention with no caveats.
  • Pushes the entire defense to the customer's system prompt.
  • Cannot name any specific mitigation mechanism.

2. Describe your internal red-teaming program for AI safety, including team composition, cadence, and scope of attacks tested.

Why it matters. Red-teaming is how a vendor looks for AI-specific failure modes before release. The buyer needs to know who does it, how often, and what it covers.

3. Do you operate a public vulnerability-disclosure program that explicitly accepts AI-specific vulnerability reports (e.g., prompt injection, jailbreaks, harmful outputs)?

Why it matters. A security VDP may accept only security vulnerabilities and turn away other AI-specific reports. OpenAI, for example, opened a separate Safety Bug Bounty in March 2026 for abuse and safety reports that do not meet its security program's criteria. A public channel that names AI issues in scope tells researchers and customers where to report them.

What the module covers

  • Prompt injection & untrusted input (17)
  • Red-teaming & adversarial evaluation (14)
  • Vulnerability disclosure & advisories (14)
  • Abuse and misuse mitigations (24)

This module is one of 7 in AI Governance. All 7 in one workbook: $499

Questions about this module

How many ai safety & responsible ai questions are there?

89: 22 for the RFI stage, 35 for the RFP and 32 deep-dive questions for the finalists.

What comes with each question?

Why it matters, what a good answer looks like, the red flags, follow-up questions, the response format, whether most buyers treat it as mandatory, and a suggested weight for scoring.

Which license do I need?

The Enterprise License covers any number of evaluations inside one organization. The Consultancy License covers use with any number of clients. Neither allows reselling or republishing the questions.

Included in these RFP packages

Each package carries this module's screening questions next to its questions about the software.

Related

Keep reading

Buyer Guide
AI Runtime Security & LLM Guardrails
Free form and approval path
IT Standards Exception Request Form

A one-form request for an exception to a technology standard, with risk, a time limit and an approval path.

State of Play
Data governance platforms

The annual review of this market: what changed, with answers from the vendors in it.