Available now: Delivered by TrendLab following a written runbook, with completed deliveries on record.Free
Free Risk Snapshot
A polite outside-in test of your public chatbot, with hand-reviewed findings.
The problem
You do not know what your bot promises, leaks or does when someone pushes it.
What we do
- About 40 to 58 tests through your public chat window, in one session
- Commitments (refunds, discounts, prices) checked against your published policies
- Prompt injection, leaked instructions, unsafe links, off-topic abuse and hand-off to a person
- Every failed, pending or errored result reviewed by a person before anything is sent
What you receive
- A short findings report or summary within 48 hours of confirming scope
- An optional 15-minute walkthrough
What you provide
- The public URL of your bot
- Confirmation that you are authorized to request testing
Limits
- A limited sample of one session, at most 60 messages
- No repeat trials, no inside-the-system checks
- Details of tests withheld from the report are not disclosed
Results describe behavior within the tested scope and configuration. They give you independent evidence for security reviews, audits and compliance work; certification itself is issued by accredited bodies.
Availability and independence
Available now: Delivered by TrendLab following a written runbook, with completed deliveries on record. Delivered by TrendLab following a written runbook, with completed deliveries on record.
Independent evaluation: we had no hand in building or advising on your agent, and our price is fixed before testing starts, so the results don't change what we're paid.
What is covered
The test areas this service always includes, with their status at the access level it uses.
- Unauthorized commitmentsCan my bot promise something we don't offer?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Invented discounts, coupons, price matchesCan it give away margin?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Contradicting published policyDoes it say the same thing as our Terms?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Wrong prices, plans or termsDoes it quote the right price?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Accepting forged "earlier agreements"Can a customer fake a promise?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Invented facts against facts supplied to the testDoes it make things up?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Self-contradiction within a replyDoes it contradict itself?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Direct prompt injectionCan one message hijack it?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Indirect injectionCan content it reads hijack it?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- System-prompt and configuration leakageCan people see our instructions?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Other customers' data in repliesCan it reveal someone else's data?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Unsafe links, markdown-image exfiltrationCan it leak data through links?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Jailbreaks and harmful outputCan it be made to say awful things?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Off-scope compute abuseIs it used as a free general AI?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Unsafe data collectionDoes it ask for passwords or card numbers?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Hand-off to a human : works, offered when needed, not overusedCan customers reach a person, and only when they need one?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Empathy and toneDoes it handle upset customers well?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Competitor handlingWhat does it say about competitors?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Abuse and harassment handlingWhat happens with hostile users?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Availability (widget loads, reply arrives)Does it actually answer?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Routing (sales / support / human)Do conversations go to the right place?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Held-out tests withheld from deliverablesCan fixes be aimed at your tests?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Confidence rating and stated limitsHow much can we rely on this?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Tamper-evident, timestamped evidenceCan you show results weren't changed afterward?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Mapping to the OWASP Top 10 for LLM Applications 2025How does this relate to OWASP?Available now: Built into the TrendLab evaluation engine and used in sealed runs.
- Mapping to the NIST AI RMF 1.0How does this relate to NIST?Available now: Built into the TrendLab evaluation engine and used in sealed runs.