Independent testing
- Independent challenge
- Source verification
- Claim-to-source testing
- Entity and identity resolution
- Provenance testing
Academic & Research Institutions
Independent evidence assurance for universities and research institutions
Keep the institutional process. Independently test the difficult evidence question.
When an integrity finding, research-security assessment, AI-assisted research product, scientific claim or institutional report may influence a consequential decision, test whether the underlying evidence actually supports it.
Aperture Research Works independently tests defined evidence questions and provides a written finding showing what is supported, contradicted, limited or not yet established.
Reality before relianceVerification before interpretationEvidence before narrative
The commercial position
Keep the institutional process in-house. Outsource the difficult evidence question.
Your Research Integrity Officer, Research Security Officer, committee, counsel and institutional leadership retain the process and the decision. Aperture can take one bounded evidence workstream when a material fact, source chain, citation, affiliation, chronology or provenance question requires deeper independent verification.
You do not need to outsource the investigation. You can independently test the one evidence question the decision depends on.
Your institution already owns the decision. Aperture independently tests whether the evidence is strong enough to support it.
Why commission an independent evidence review?
The institution is buying separation, challenge and a written evidentiary view.
The value of an independent review is not that an external reviewer knows the institution better than its own team. The value is separation. Aperture can examine one defined evidence question without being responsible for defending the original research, protecting a preferred conclusion or making the institution's final determination.
Evidence discipline
- Source-independence analysis
- Contradiction identification
- Material-gap identification
- Chronology reconstruction
- A documented evidence trail
Institutional value
- A bounded outside resource
- A narrow scope rather than a broad consultancy assignment
- A human-approved written product
- A record of what is and is not supported
- Clear limits on permitted reliance
When an external evidence workstream earns its place
Six situations in which a bounded independent review can add practical value.
Aperture does not assume that the institution lacks a process or capable people. The fit arises when one part of the existing process benefits from separation, capacity or deeper evidence work.
A difficult question inside an active matter
One citation, data point, chronology, source, document or factual proposition requires deeper testing than the wider team can reasonably absorb.
Independent expertise or conflict management
A separate reviewer adds useful independence or specialised source, entity, provenance or contradiction research.
Institutional-record readiness
A material finding must be traceable to the supporting, limiting and contrary evidence before the record carries stronger reliance.
AI-related evidence verification
A detector, citation or AI-assisted output has raised a concern. The underlying proposition now needs to be tested against the evidence.
Research-security factual verification
An existing security programme has identified one entity, affiliation, funding link, collaboration or chronology that needs deeper factual resolution.
Overflow and deadline capacity
The authorised institutional team retains control while Aperture handles one clearly bounded, labour-intensive evidence workstream.
The problem is not always a lack of research
The problem can arise after substantial research and specialist work already exist.
A material conclusion has been reached. A report has been drafted. An allegation has been assessed. An AI-assisted research product has been produced. A foreign collaboration has been reviewed. A scientific claim has been documented. A mitigation decision is being considered.
At that point the question changes. The institution no longer needs more information for its own sake. It needs to know whether the existing evidence can responsibly carry the conclusion.
This is the point at which Aperture fits.
Current context verified 1 September 2026
NSPM-33 NSF research-security updates Revised 42 CFR Part 93 ORI generative AI guidance Thirty-institution security review
Why this matters now
Three overlapping processes create different evidence questions. The 31 August security-review deadline has now passed.
The federal research-security framework began strengthening protections for government-supported research.
Additional assessment, documentation, training and certification requirements took effect for NSF-funded activity.
The revised requirements became applicable to covered allegations received on or after this date.
ORI addressed generative AI in misconduct proceedings and confirmed that AI detectors alone do not meet the evidentiary burden.
Thirty institutions received federal research-security review notices with findings and actions due by 31 August.
The ongoing integrity and AI context
The revised Public Health Service research-misconduct regime is already active for covered new allegations. ORI's generative AI guidance does not create a separate 31 August deadline. It reinforces an ongoing evidentiary question: a misconduct finding must be supported by evidence beyond the use of AI detectors alone.
ORI's September 2026 RIO Boot Camp also addresses the 2024 rule, case management and RIO responsibilities. This is ongoing implementation and professional practice, not an expired one-day market.
Primary sources: HHS final rule, ORI generative AI guidance and ORI's September 2026 RIO Boot Camp notice.
The dated 30-institution event
On 17 August 2026, the Department of War announced formal notifications to 30 domestic academic institutions. The notified institutions were directed to review identified foreign collaborations and report findings and actions by 31 August 2026. That reporting deadline has now passed; any later independent review concerns the evidentiary quality, traceability and defensibility of the record produced, not a claim that a second audit is inevitable.
That date applies to this specific review. It is not the end of the wider research-integrity, AI or research-security evidence problem.
Primary source: Department of War release, 17 August 2026.
Post-submission evidence assurance after 31 August 2026
Following the August 2026 federal research-security review process, affected institutions may have an established methodology, evidence record, findings and mitigation conclusions. Aperture can independently test whether that record supports the conclusions that may subsequently be relied upon.
There is no claim that every institution will face a second audit. The legitimate assurance question is narrower: can the submitted record traceably support later assessment, mitigation or funding-related reliance?
Additional official context: NSF Important Notice No. 149 and ORI's March 2026 guidance release. Aperture is not affiliated with or endorsed by ORI, HHS, NSF, the Department of War or any university.
What are you responsible for?
Four university buying situations. One independent evidence function.
Research Integrity Officer, research compliance, research leadership, counsel where appropriate
Research integrity and misconduct evidence
Problem: A factual proposition in an assessment, inquiry or investigation needs independent evidentiary testing.
Aperture role: Test one defined proposition against the available record.
- Does cited literature support the allegation?
- Can a disputed reference be authenticated?
- Does the research record support the factual assertion or chronology?
- Are citations independent, or repetitions of one source?
- What evidence weakens the leading interpretation?
Natural first application: Research Integrity Evidence Review.
Boundary: Aperture does not make the institution's misconduct determination.
Research Security Officer, Export Control Director, research compliance, research leadership
Research security and foreign-collaboration evidence
Problem: The institution needs to establish what the evidence shows about a person, entity, affiliation, appointment, funding relationship or collaboration.
- Are identities, aliases and translations resolved?
- Is co-authorship being confused with affiliation?
- Can an appointment or funding relationship be established?
- Are sources genuinely independent?
- Does the evidence support the exposure or mitigation conclusion?
Natural first application: Research Security Evidence Review or Post-Submission Research Security Evidence Review.
Boundary: Aperture does not presume a foreign collaboration is improper, provide export-control legal advice or make government eligibility determinations.
Research integrity, research administration, programme leadership, scientific teams, report owners
Generative AI, citations and research products
Problem: An AI-assisted output looks plausible, but its factual foundation has not been independently tested.
- Does each source exist and contain the stated proposition?
- Is the source authoritative for that proposition?
- Are citations circular or dependent?
- Has generated synthesis made the language stronger than the evidence permits?
- Can material passages be traced to original evidence?
Natural first application: AI Research and Citation Evidence Review.
Boundary: AI detection is not the product. Evidence verification is the product.
Scientific programme leaders, Principal Investigators, research infrastructure and AI-assurance teams
Scientific, technical and AI-assurance evidence
Problem: A scientific, methodological, dataset or AI-related claim needs independent challenge before consequential reliance.
- Does the literature support the claim?
- Can the methodology or dataset provenance be reconstructed?
- Do documented tests establish the reliability claim?
- Are limitations, adverse findings, endpoints and denominators visible?
- Does documentation demonstrate reproducibility?
Natural first application: Scientific Evidence and Methodology Review or AI Assurance Evidence Review.
Boundary: Aperture does not certify scientific validity beyond the evidence reviewed or replace qualified specialists.
AI detection is not evidence assurance
A signal may identify a question. It does not prove the answer.
Aperture does not compete with the institution's detection tools. We test the proposition behind the signal. Does the source exist? What does it actually say? Can the passage, claim or data point be traced? Are the sources independent? What remains unproved?
The client is buying the evidence test, not the detector.
Where Aperture fits
The institution retains the process, authority and final responsibility.
Within a research-integrity process
- Allegation or evidence question
- Institutional assessment
- Defined factual proposition
- Aperture independent evidence test
- Written evidence finding
- Institutional process continues
- Institution retains final responsibility
Within a research-security process
- Identified relationship or concern
- Institution defines the question
- Aperture maps the relevant evidence
- Identity, affiliation, funding and chronology test
- Contradictions and gaps
- Written evidence finding
- Institution decides what action is appropriate
Aperture may test a specific factual or evidentiary issue before, during or after an authorised institutional process. We do not control the process or decide the outcome. We supply a separate evidence product into it.
The current 42 CFR Part 93 definition of an institutional member can include subject-matter experts and consultants affiliated by contract or agreement. Section 93.305 also addresses a committee, consortium or other person acting on an institution's behalf, subject to conflict and compliance controls. This does not mean ORI approves Aperture or authorises universities to transfer their formal responsibilities. Read 42 CFR 93.219 and 42 CFR 93.305.
Three engagement points
Commission the evidence test at the stage where it adds value.
Prevent unsupported conclusions from entering the decision record.
Use when a finding, report or assessment is still being developed.
Isolate one disputed or resource-intensive factual question.
The institution keeps control of the wider process.
Test whether the completed methodology and evidence chain support the conclusions.
This is the relevant framing for the 30-institution security event after 31 August 2026.
Three clear ways to engage
Start with the smallest scope that can answer the real decision question.
The normal route is one Focused Evidence Review first. A larger audit or standing supplier arrangement follows only when the institution has seen the work and a further need is real.
Research Reliance Review
You already have the research. Now test whether it is ready to carry the decision.
A Research Reliance Review examines whether an existing research product is sufficiently traceable, supported, challenged and complete for the decision it is expected to inform. It can be commissioned as a Focused Evidence Review for one proposition or as a Claims & Evidence Audit for a larger record.
Focused Evidence Review
Typical limited academic engagement: US$1,500 to US$3,000
One material question. One defined scope. One written finding.
- One disputed proposition
- One citation chain or research record question
- One foreign affiliation or funding relationship
- One methodological, dataset or scientific claim
- One section of an institutional report
Final pricing depends on evidence volume, complexity, urgency, intended use, jurisdiction, specialist dependencies and information-handling requirements.
Define the Evidence QuestionClaims & Evidence Audit
Typical bounded academic engagement: US$3,000 to US$7,500
Use when several material propositions or a larger report require systematic evidence testing.
- Supporting and contradictory evidence
- Source authority and independence
- Material gaps and uncertainty
- Permitted reliance
Standing Evidence Assurance Arrangement
Complete supplier onboarding once, agree the commercial and information-handling framework, and commission separate bounded matters as evidence questions arise.
- Master Services Agreement where appropriate
- Matter-specific Statements of Work
- Purchase orders where required
- Separate written scope for each assignment
- Conflict and capacity checks for every matter
Matter-Specific Evidence Support
Where an active matter contains several defined research tasks, Aperture may provide multiple evidence reviews, chronology reconstruction, entity mapping, source acquisition, citation validation, evidence matrices, contradiction review and ongoing written findings. Pricing is quoted by scope. No universal price is implied.
Optional Priority Evidence Review Access
For larger institutions, a standing agreement may allow individual work orders when an urgent evidence question arises. This is not automatically a monthly retainer. Its commercial value is reduced contracting friction after initial onboarding.
You do not need to outsource the matter
You can outsource one evidence question.
Begin with the smallest proposition whose answer could materially change the decision.
- Did this source actually support the claim?
- Can this affiliation be independently established?
- Does this research record support the stated chronology?
- Are these three citations genuinely independent?
- Can this dataset's provenance be reconstructed?
- Does this evidence support the stated mitigation conclusion?
No further work begins automatically. If the first review identifies a larger evidence problem, Aperture provides a separate written scope. The institution decides whether expansion is proportionate.
What the institution receives
A usable written decision record, not a research dump.
According to scope, the record may include
- Executive finding
- Central Question
- Defined proposition
- Scope and exclusions
- Source register
- Evidence map
- Claim-to-source matrix
- Chronology
- Entity-resolution record
- Contradiction register
- Evidence-gap register
- Provenance observations
- Specialist dependencies
- Limitations
- Reliance finding
- Proportionate next evidence steps
The Aperture reliance finding
SUPPORTED
The evidence reviewed materially supports the proposition within scope.
PARTIALLY SUPPORTED
Important elements are supported, but limitations prevent stronger reliance.
NOT SUPPORTED
The evidence reviewed does not establish the proposition.
NOT YET ESTABLISHED
The evidence available within scope does not permit a responsible determination.
These are Aperture evidence-assurance classifications. They are not legal burdens of proof and do not replace any standard applicable to the institution's own proceeding.
Why a separate outside review?
Aperture adds a defined function beside the institution's existing capability.
Why not only internal staff?
Internal teams know the institution, research environment and applicable process. Aperture adds separation, extra research capacity and a written independent view for one defined question. It does not imply internal personnel are incapable.
Why not only AI?
AI can accelerate discovery and analysis. It cannot establish a material fact merely by producing an answer. The client is buying an independently tested evidence record, not access to AI.
Why not a broad consulting engagement?
Aperture is useful when the institution needs one difficult proposition tested rather than a large transformation programme. The scope can remain small, precise and decision-specific.
A repeatable institutional route
Build the supplier framework once. Keep every evidence matter separately scoped.
Supplier onboardingConfidentiality and security termsServices agreement where appropriateMatter-specific Statement of WorkPurchase order where requiredWritten matter reference and scope
Aperture accepts international matters subject to jurisdiction, lawful-use, professional-boundary and information-handling review. Where supplier onboarding, contracting or tax documentation is required, it can be handled through the institution's normal procurement process after the proposed scope is accepted.
No standing arrangement guarantees acceptance of every matter. Independence, conflicts, scope, lawfulness, specialist requirements and capacity are checked for each assignment.
Working beside the authorised institutional team
Clear roles protect the institution and the independence of the evidence review.
Working alongside the RIO
The Research Integrity Officer retains the institution's formal responsibilities. Aperture can be assigned a bounded factual proposition, source question, chronology issue, citation review or other evidence task.
Working alongside the Research Security Officer
The institutional officer retains responsibility for the research-security programme and decisions. Aperture can test defined factual questions involving entities, affiliations, appointments, funding, publications, chronology and source provenance.
Working alongside counsel
Aperture can perform independent factual or evidence research under counsel's direction. Aperture does not provide legal advice. Any privilege or work-product structure must be determined by the institution and its counsel. Aperture does not promise privilege.
Trust, independence and limits
The preferred conclusion is not the product. The tested evidence record is.
Aperture cannot promise
- That the preferred institutional conclusion will be confirmed
- That an allegation will be sustained or rejected
- That a funding body or agency will accept a conclusion
- That ORI will accept an institutional finding
- That an AI system is safe
- That every evidence gap can be resolved
Aperture will test the evidence and make its limits visible.
Before a matter is accepted
- Client, matter and relevant entity conflict checks
- Lawful-use and jurisdiction assessment
- Professional-boundary review
- Information-handling review
- Specialist dependency review
- Capacity assessment
Aperture may decline a matter if independence cannot be maintained.
What Aperture does not do
We do not determine academic guilt or innocence, replace a Research Integrity Officer or institutional committee, provide legal or export-control advice, certify regulatory compliance, provide classified threat assessment, make government funding or eligibility determinations, certify AI models as safe or replace qualified scientific and technical specialists. We do not treat an AI detector result as proof.
Start without sending the case file
Most matters can first be scoped from a sanitised description of the evidence question.
What the initial enquiry should contain
- The institution and responsible role
- The type of matter
- The defined evidence question
- The decision or finding affected
- General evidence volume
- A real deadline
- Known information restrictions
Do not initially send
- Classified information or Controlled Unclassified Information
- Export-controlled technical data
- Protected health information or restricted student records
- Privileged legal material
- Sensitive personal data or restricted research records
- Passwords, credentials or non-public source code
See the method applied to evidence failure
Relevant independent public-source examples
These examples demonstrate Aperture methods. They are not represented as university client engagements.
Mata v. Avianca
Generated or defective citations, source authentication and human gatekeeping.
See the evidence methodProstate treatment claim
Scientific claim verification, qualification and specialist boundaries.
See the evidence methodElizondo v. City of Laredo
Claims, records, contradictions and the limits of an evidence chain.
See the evidence methodAcademic FAQ
Independent evidence review and the institution's own determination remain separate.
Why would we use Aperture if we already have internal research and compliance teams?
Your internal teams retain institutional knowledge, authority and responsibility. Aperture provides a separate evidence-assurance function for a defined question. That adds independent challenge, additional research capacity and a written evidentiary view without outsourcing the institutional decision.
What is the smallest engagement?
One material evidence question. A Focused Evidence Review can test a single proposition, source chain, affiliation, citation issue, methodology question or representative evidence sample. The institution receives a written finding and can stop there.
What happens if the first review finds a bigger problem?
Nothing expands automatically. Aperture identifies the additional evidence questions and provides a separate written scope. The institution decides whether further work is proportionate.
Does Aperture replace our Research Integrity Officer?
No. The RIO retains the institution's formal role. Aperture may provide a bounded independent evidence product within the authorised process.
Can Aperture investigate a single evidentiary proposition?
Yes. The normal starting point is one narrow proposition that can materially affect a decision.
Can Aperture review evidence during a research-misconduct matter?
Potentially, where the institution has lawful authority to share the material and its policies and applicable requirements permit the external task. Aperture does not make the institutional determination.
Can Aperture work under the direction of university counsel?
Yes, for independent factual or evidence research. Counsel must determine any privilege or work-product structure. Aperture does not provide legal advice or promise privilege.
Can Aperture determine whether misconduct occurred?
No. Aperture can state what the evidence supports, does not support or has not yet established. The formal determination remains with the institution.
Can Aperture verify AI-generated citations?
Yes. The review can test whether references exist, whether their bibliographic details are authentic and whether each source supports the attached claim.
Do you rely on AI detectors?
No. A detector may provide a signal, but it does not establish a misconduct finding or prove a factual proposition. Aperture tests the underlying evidence.
Can Aperture review foreign affiliations and research collaborations?
Yes, within lawful and agreed sources. Aperture does not presume that a foreign relationship is improper merely because it exists.
Does Aperture provide export-control advice?
No. Aperture can test factual evidence relevant to an affiliation, relationship or chronology. Export-control advice remains with qualified counsel and institutional professionals.
Can Aperture review an already completed research-security assessment?
Yes. A post-submission review can test the methodology, traceability, entity identification, source relationships, contradictions, omissions and evidentiary basis of the conclusion.
Can Aperture review only one part of a report?
Yes. One decision-relevant section or proposition can be scoped independently.
What will we receive?
A written record of the question, scope, evidence examined, sources, material support, contradictions, gaps, limitations and the reliance classification reached within scope.
Can we start without sharing confidential material?
Yes. Initial scoping can usually begin from a non-sensitive description, sanitised methodology summary or authorised representative sample.
Can we use Aperture for recurring matters?
Yes. After supplier onboarding, separate bounded matters can be commissioned under their own written scopes, subject to conflict, capacity and acceptance controls.
How does a standing services arrangement work?
The institution and Aperture may agree a general commercial and information-handling framework. Each assignment still requires a separate Statement of Work or equivalent scope and any required purchase order.
Can a US university engage Aperture as an international supplier?
Potentially, yes. Acceptance depends on jurisdiction, lawful scope, procurement, information handling, sanctions, conflicts, capacity and specialist requirements.
Does Aperture guarantee a preferred conclusion?
No. The fee purchases independent testing, not confirmation. Supporting and weakening evidence remain visible.
Does Aperture investigate research misconduct?
Aperture does not replace the institution's research-misconduct process. Where lawful and permitted, we can independently test a bounded factual or evidentiary question and provide that written evidence product to the authorised institutional process.
Can Aperture review scientific methodology?
Yes, within a defined evidence scope and subject to professional boundaries. We can test traceability, source-to-claim alignment, documented assumptions, calculations, provenance, contradictions and limitations. Specialist scientific judgement remains with appropriately qualified experts.
Can Aperture work alongside a Research Security Officer?
Yes. The Research Security Officer retains institutional responsibility. Aperture can independently test defined factual questions involving entities, affiliations, appointments, funding relationships, publications, chronology and source provenance.
Can Aperture certify regulatory compliance?
No. Aperture provides an evidence-assurance finding within an agreed scope. It is not a regulatory certification, legal opinion or substitute for the institution's own compliance determination.
Can Aperture certify an AI system?
No. We may test whether specified documentation, provenance, testing and observed results support a defined reliability proposition. We do not certify an AI system as safe, compliant or suitable for every use.
Start with one question
Your institution does not have to outsource its decision. It can independently test the evidence the decision depends on.
Aperture Research Works takes one defined material question, traces it to the underlying record, tests the sources, challenges the leading interpretation, identifies contradictions and gaps, and returns a written finding showing what the evidence can responsibly support.
Start with one question. Expand only if the evidence justifies it.
Reality before reliance.
Research in the AI era
AI CAN ACCELERATE ACADEMIC RESEARCH. IT CANNOT TURN AN UNSUPPORTED PROPOSITION INTO EVIDENCE.
Aperture can independently test material citations, source lineage, entity or dataset attachment, chronology, contradictions and evidential gaps where a university or research institution needs a separate written evidentiary view.
The institution retains academic authority, disciplinary judgement and final responsibility.