Declaring Conflicts of Interest and Conflicts of Commitment: What You Disclose, and How Far It Goes

Hello, this is Ryuta Hamamoto from TIMEWELL.

"Conflict of interest declarations? We do those every year."

Whenever I talk about research integrity at a university, that sentence comes back at me. Then I ask what the declarations actually contain, and at a fair number of institutions the answer stops at an annual form covering corporate directorships and holdings of unlisted shares. A perfectly respectable conflict of interest process, and a good deal narrower than what the Japanese government is now asking institutions to disclose.

This piece works through the disclosure practice that sits underneath research integrity. What separates a conflict of interest from a conflict of commitment, what you declare and how far it reaches, what you register in e-Rad, what institutions are expected to build, and what happens when something is left off. For the shape of the regime as a whole, see the complete guide to research integrity and research security.

Conflict of interest and conflict of commitment are two different things

Let me get the vocabulary straight first, because designing a declaration form while these two are blurred together produces a form that does not work.

Before that, the two terms above them. Research integrity means the efforts that keep research activity sound and honest. The procedures manual for ensuring research security, published by a Cabinet Office expert panel in December 2025, positions it as something that "covers all research activity." Research security, its counterpart, protects research from risks such as technology leakage, and is deliberately narrower: it "covers research activity that the state or the research institution has judged should be protected"¹. MEXT frames the sequence the same way, with research integrity as the foundation and research security built on top². COI and COC declarations sit at the very bottom of that foundation.

The origin of the distinction is the report of the Conflict of Interest Working Group established under MEXT's Council for Science and Technology, published in November 2002. That report divided conflict of interest in the broad sense into two. Conflict of interest in the narrow sense describes a situation where benefits that a faculty member or the university obtains through industry-academia-government collaboration (licensing income, remuneration from secondary employment, unlisted shares and so on) collide with the university's responsibilities in teaching and research. Conflict of commitment describes a situation where a faculty member, mainly through secondary employment, owes a duty of performance to a company that cannot be reconciled with their duty of performance at the university³.

The report explains the difference as a difference in the causal factor. In both cases what is at stake is the discharge of a university responsibility; the distinction is whether the factor is a benefit received from a company or a duty owed to it³. Put roughly, conflict of interest is about money and equity, conflict of commitment is about the allocation of time and effort. That reading will carry you through most of the operational work.

One more passage in that report gets to the heart of this field. The concept of conflict of interest does not itself denote a failure to discharge university responsibilities. The problem is less the state of affairs than the university's indifference to it, which makes it look, to the public, as though those responsibilities were not being discharged, putting trust in the university at risk³. Being in a conflicted position is not a wrong. Leaving it unattended is.

In the research integrity context the definition then shifts. The policy adopted by the Integrated Innovation Strategy Promotion Council on 27 April 2021 treats the two as a single bundle: a situation where benefits obtained by a researcher or institution through research activity collide with the responsibilities required in research carried out using national or funding agency money, or with the responsibilities each institution requires of its researchers⁴. So there is a precise definition built for the industry-academia context, and on top of it one built for the risks of internationalisation and openness. That two-storey structure keeps you from getting lost in your own institution's rules.

This area also runs continuously into security export control. If you want to know roughly where your own export control arrangements stand, our free export compliance self-check takes a few minutes.

What you declare, in three layers

So what do you actually hand over? Reading the 2021 policy together with the Guidelines on the Proper Execution of Competitive Research Funding, as revised in response to it, the information requested falls into three layers.

Layer	Information covered	How it is submitted
1	All current applications for and receipts of research funding, not limited to domestic competitive funding: subsidies, grants, joint and commissioned research funding, including funding from outside Japan (programme name, project title, period, budget, effort)	Stated in the application documents or in e-Rad
2	All current affiliations and titles, including secondary employment, participation in foreign talent recruitment programmes, and unpaid positions such as emeritus professorships with no employment contract	Stated in the application documents or in e-Rad
3	Information needed to ensure transparency across all research activity the person is involved in, including donations and support in the form of facilities and equipment rather than money	A declaration that it has been properly reported to the institution

The people covered are the principal investigator and the co-investigators on the project being applied for⁴⁵. A word on "effort," in the first layer: the percentage of a researcher's total annual working time devoted to a given project, taking the total as 100%. You declare the allocation of your time as a number, which connects it directly to conflict of commitment further down.

In practice, the layer that trips people up is overwhelmingly the third, and it is built differently. Layers one and two are information you write out each time you apply. Layer three is a declaration that you have "duly reported this to your institution." Turn that around: if the institution has no mechanism for receiving those reports, the researcher has nothing to declare. That is precisely why the regime asks institutions to put rules in place.

The contents of layer three are broader than most people assume. The policy defines support other than research funding as including the free provision of research facilities, equipment, instruments and other goods, and the free provision of services⁴. Not a single yen needs to have moved. Being allowed to use an instrument at no charge. Having analysis or sample preparation done for you as a favour. Neither is unusual in a working lab, and my sense is that the understanding that these are disclosable has not spread very far.

The Cabinet Office's model checklist adds a note on the scope of "remuneration": grants, salary from concurrent posts, prize money, gifts, donations, travel expenses, speaking fees, writing fees and the like⁶. The inclusion of gifts, travel expenses and speaking fees is the part worth flagging. You were invited to speak at a university overseas, and they covered your airfare and paid you a fee. That can be reportable.

There is also relief where a non-disclosure agreement prevents you from writing something down. So as not to chill industry-academia collaboration, the guidelines limit the information required to what is necessary to confirm whether adequate effort can be secured: as a general rule, only the partner institution's name, the amount received, and effort. And for the time being, where an executed non-disclosure agreement makes submission difficult, the partner's name and the amount may be omitted⁵. JST's e-Rad input design follows the same logic: set the non-disclosure flag to "yes" and everything other than effort becomes unnecessary⁷. There is a route that satisfies the disclosure obligation without breaching the contract.

Why layer two includes "emeritus professorships with no employment contract" makes sense once you look at a real case. A collection of near-miss cases produced under a MEXT commissioned project records a professor emeritus, three years past retirement, with no employment relationship and no involvement in the administration of research funds, who was approached by a talent recruitment programme: five years of education and research at a university overseas, exceptional funding and state-of-the-art facilities, all results and technical information assigned to that university, at roughly three times their salary while in post. They brought it to the administrative office⁸. "No employment relationship, so it is not the institution's business" does not hold in every situation.

The same collection has a clean illustration on the conflict of commitment side. A professor was invited to take up an adjunct lectureship overseas. HR compared the letter of appointment against the draft contract and found a gap. The letter described roughly two hours a month including on-site attendance. The draft contract set annual working time at over a month, required on-site attendance at a technical exchange meeting twice a year, and made publication of multiple papers and joint application to an international programme obligatory. The professor asked for the draft to be revised⁸. Conflict of commitment, boiled down, is that discrepancy in hours. How many you wrote on the form, against how many it will actually take. You need something that checks whether the two agree.

One unchecked box in e-Rad and you cannot apply

Let me bring the policy discussion down to the screen in front of you. This disclosure is implemented in the cross-ministerial R&D management system, e-Rad, which since 15 March 2022 has included research integrity fields in researcher records. Log in, open the dropdown under your name at the top right, go to view and edit researcher information, and open the tab for your affiliated institution. Three fields sit there: research funding outside e-Rad, secondary employment and participation in foreign talent recruitment programmes and unpaid positions such as emeritus professorships, and declaration status⁷.

"Research funding outside e-Rad" is an odd phrase, so let me unpack it. It means funding that is not competitive research funding, plus competitive funding that was not applied for through e-Rad. Funding already registered in e-Rad as ongoing, applied for, or due to be received does not need to be entered again⁷. Grants from overseas foundations, joint research funding from companies, and internal university funds are the typical entries.

The last field is a declaration status checkbox. Even with nothing at all to enter in the two fields above it, you still have to tick it⁷. This is where accidents most often happen. JST's input manual states that if a participating researcher has not registered their declaration, the application throws the error "you cannot apply because a researcher who has not made the research integrity declaration is participating."⁷ For CREST it is mandatory for every main co-investigator, not just the PI.

KAKENHI works the same way. JSPS's FY2026 application guidelines note that since the FY2025 call, research integrity information registered in e-Rad has been transferred into the electronic application system, and state plainly that where the principal investigator or a co-investigator has not registered their declaration status, the application cannot be submitted. They warn too that the transfer normally takes 30 to 60 minutes and can take several hours at busy times, so registering on the day of the deadline is risky⁹. One co-investigator forgetting one checkbox and the entire proposal cannot be filed. That point is worth repeating at every internal briefing.

The number of programmes in scope has changed a great deal as well. According to Cabinet Office materials, a revision on 17 December 2021 extended coverage from the former competitive funds to all open-call research funding programmes, taking the count from 20 to 100, applied progressively to calls issued from April 2022 onwards¹⁰. This is no longer a matter for a handful of large programmes.

What happens when a declaration is inaccurate

This is the part I want to state precisely, because the measures come in tiers. Pulling together what the guidelines and the policy provide gives the following⁴⁵.

Situation	Measure
Where "unreasonable duplication" or "excessive concentration" is found, or where application documents are confirmed to contain statements at odds with the facts	Rejection of the project, revocation of its selection, or reduction of its allocation
Where it emerges that, contrary to the declaration, proper reporting to the institution was not carried out	The same (rejection, revocation, reduction)
Where funds are confirmed to have been received by deception or other wrongful means	Return of the funds, restriction of eligibility for the competitive funding concerned, and restriction on applying to other competitive funding including that of other ministries

The application ban for wrongful receipt runs, as a general rule, for five years from the fiscal year following the year in which the subsidy was returned⁵. For comparison, misuse of research funds carries ten years where funds were diverted for personal gain, five years where the impact on society is large and the conduct is judged seriously improper, two to four years otherwise, and one year where the impact and the impropriety are both judged small; fabrication, falsification or plagiarism carries two to ten years⁵. Where the impact and the impropriety are small and the amount misused is minor, the case may end in a strict warning with no ban at all.

There is one point where I would avoid stating things too firmly. The procedures manual published by the Cabinet Office expert panel in December 2025 describes conduct that breaches it in the future tense: the guidelines could be revised so that such conduct is treated as wrongful receipt and measures such as application bans "could be" imposed¹. The current published text of the guidelines, as revised on 17 December 2021, makes no reference to the manual. A regime under which breaching the manual immediately triggers a five-year ban is not yet in operation, and that is the accurate reading as of July 2026.

The scope of application invites misunderstanding too. The manual applies to calls designated as "Specified Research and Development Programs": competitive funding programmes premised on open publication of results, which the responsible ministry, with the funding agency, has designated as potentially involving technologies on the list of critical technology areas. In JST's FY2026 guidelines, only five CREST research areas within the Strategic Basic Research Programs are designated; PRESTO and ACT-X are outside scope. Application begins with newly selected FY2026 projects, and the institutions covered are those described in the commissioned research contract as "universities and the like" and "companies and the like." The Acquisition, Technology and Logistics Agency's Innovative Science and Technology Initiative for Security is another, from its FY2026 call, which runs from 13 March to noon on 20 May 2026. This does not apply uniformly to every call.

The verification work is not unbounded either. The manual sets out 13 due diligence items: academic background, research and employment history, research funding received, support other than research funding received, whether the person is first, corresponding or co-author on published papers, patent filings, participation in foreign talent recruitment programmes, and so on. Items 3 to 8, running from research funding received to any measures imposed under the guidelines, plus item 10 on affiliation with a listed institution and item 11 on relationships with researchers at a listed institution, cover the past three years including the year of application. "Relationship" in item 11 is itself narrowed to concrete facts: conducting joint or commissioned research, writing and publishing a co-authored paper, or appearing as a named co-presenter in an oral presentation at a conference. Those covered are the PI and Co-PIs plus research participants at the lead institution, students included. The manual then says the verification may be done using "self-declared information, open source information and other information that each lead institution can ordinarily obtain"¹. Nobody is being asked to run a private investigation. The practical side is in How to run due diligence and how to answer the questionnaire.

And there is something that bites earlier than any formal measure. The government says so itself, in a note to the checklist: where a researcher applying to a call fails to submit information about joint research with a foreign institution that is subject to restrictions abroad, they risk being suspected of false statement or conflict of interest, and their personal credibility suffers⁶. Trust goes before sanctions do. A Cabinet Office overview cites, as inappropriate cases identified in the United States, false declarations about involvement in talent recruitment programmes and non-disclosure of founder and major shareholder status in a foreign company, at the stage of indictment or resignation¹⁰. None had reached a final judicial determination as of that document's date, so I will not pass judgment on the individual cases. The pattern still shows that non-disclosure itself can attach directly to a career.

The institutional homework has moved from writing rules to verifying claims

The regime does not put all of this on researchers. The 2021 policy asks institutions to receive and keep updated the information needed for personnel and organisational risk management (employment and research history, affiliations and titles, research funding and non-funding support from outside the institution and its sources), to put in place rules and management arrangements covering conflict of interest and conflict of commitment, and to conduct risk management on that basis as they would for industry-academia collaboration⁴. The guidelines add that funding agencies will state in their calls that they may ask an institution about the state of its rules and how it captures and manages this information⁵.

The Cabinet Office has published two model checklists for this: one for researchers dated 17 December 2021, one for universities and research institutions dated 29 June 2023. The dates differ, so do not mix them up. Both are in three parts: general matters, procedures for collaboration with foreign institutions and for contracts and remuneration, and matters relating to the counterparty⁶¹¹.

When the institutional checklist was revised in 2023, two items were added that carry real operational weight. Whether there is a mechanism for verifying information submitted by researchers and staff, such as comparison against separately obtainable information, calibrated to the level of technology leakage risk. And whether there is a mechanism for redoing the risk assessment when the risk level of an external institution someone has a relationship with changes¹¹. Read plainly, this says receiving declarations is not enough. You compare what you received against outside information, and when the situation changes you assess it again.

How far institutions have got can be checked against numbers. According to the follow-up survey summary the Cabinet Office published in December 2025, of 344 national, public and private universities and similar institutions, 86% (296) had COI and COC rules in place and 4% (15) planned to introduce them, roughly nine in ten combined. The 87 national university and inter-university research institute corporations scored 100% on every item. On the other hand, 66% had a mechanism for objectively verifying the facts behind reported information, 69% had one for dealing with high-risk situations before they materialise, and 66% had one for receiving updates, with a further 6 to 7% planning each¹².

The rules are essentially done. The verification and updating machinery has not caught up. That is how I read the picture in 2026, and the gap is not down to anyone being lazy. Checking co-authorship and shareholding relationships against public information simply does not finish if you do it by hand. The same case collection records a professor who co-authored a paper overseas on research that raised no issue whatsoever under the Foreign Exchange and Foreign Trade Act. The procedural side had been left to the counterparty, and it emerged only later that the co-author list included a researcher affiliated with an institution appearing on national control lists, which pulled in the university headquarters⁸. The professor had never met that researcher. Relationships can arise through routes you have no way of declaring yourself.

A note on how to talk about this. Appearing on a control list is a regulatory designation, not a finding about the merits of the institution or the researcher. Being listed adds one verification step to the procedure; the listing itself signifies neither wrongdoing nor military diversion of technology. The same holds for the "specific categories" in security export control, the classification introduced in May 2022 that groups residents who may be under strong influence from abroad and makes them subject to individual verification when technology is provided. METI states expressly that falling under a specific category does not mean a person is regarded as posing a security concern. Do not translate the result of a name match into a judgment about a person. Operating this without that discipline costs you the trust of the researchers you depend on.

The matching side of this work overlaps directly with what we have built at TRAFEED for export control. TRAFEED is structured around a knowledge graph of more than 200 million records, including roughly 90 million papers, roughly 100 million patents and roughly 300,000 researchers (a knowledge graph being a database where papers, patents, organisations and people are joined by relationship edges you can follow), with company lists and national sanctions lists connected to it. It is the world's first AI agent in the field of Japanese security export control, covering both list controls, which define regulated goods by specification, and catch-all controls, which require verification based on end use and end user; that is a finding from our own research as of March 2026. In joint validation work with Okayama University against roughly 30,000 past screening records, we confirmed AI determination accuracy of 95% or higher (our own research). We hold patent no. 7862062, have been adopted by more than 20 organisations, and are extending into shareholder and capital relationship research. Features aimed specifically at research integrity are still in development, but the technical continuity is real: comparing what was declared against public information and surfacing relationships that may have been omitted is the same shape of problem. The final determination, though, belongs to the research institution, or in a corporate setting to each company's export control officer. What we can take on is assembling the material and leaving it in a form that can explain, later, why a conclusion was reached.

One premise to close on, because it should never be dropped. MEXT states that efforts to ensure research security are not about pursuing zero risk or restricting research broadly, that their purpose is to move research and international collaboration soundly forward, that the principle is to mitigate risks within a reasonable scope, and that discrimination on grounds of race, nationality and the like must not occur². The manual takes the same position: what it asks for is not zero risk but reasonable handling proportionate to the degree of risk, once the technologies in scope have been narrowed. The prohibition on discriminatory treatment on grounds of nationality, race, religion, culture and the like is written into both documents¹². COI and COC declarations are not an instrument for monitoring researchers with particular attributes. They are a procedure for reaching a state where you can explain who you work with, where your support comes from, and what you are doing with it. I have made the same argument in Research integrity and economic security at universities.

What to take away

• Conflict of interest in the narrow sense is about money and equity; conflict of commitment is about the allocation of time and effort. The research integrity policy defines the two together
• Declarations come in three layers: all research funding, all affiliations and titles, and support including donations and the free provision of equipment and services. The third layer is a declaration about having reported to your institution
• Remuneration extends to gifts, travel expenses, speaking fees and writing fees. Where a non-disclosure agreement applies, the partner's name and the amount may be omitted
• Without the e-Rad declaration registered you cannot apply. Everyone including co-investigators needs it, and coverage has grown from 20 programmes to 100
• Measures come in tiers: rejection, revocation or reduced allocation for inaccurate statements; return of funds and, as a rule, a five-year ban for wrongful receipt. Measures for breaching the procedures manual are still written in the future tense
• The manual applies only to Specified Research and Development Programs. At JST that is five CREST research areas, with PRESTO and ACT-X outside scope. Verification uses information you can ordinarily obtain, and zero risk is not required
• Listing is a regulatory designation, not a judgment on the merits. Put it in writing, in advance, that a match result is not an assessment of a person
• Where institutions have got to: rules at roughly 90%, objective verification of the facts at 66%. The homework has moved from drafting to verifying

Honestly, I do not think there is much value in making the declaration form thicker. Add more boxes and, if nobody is checking whether what was written is accurate, all you have added is burden. Trust does not go up. So do not start with more fields. Take the information you already hold and run it against public sources once, end to end. You will learn more than you expect, and the gaps that surface will tell you which mechanism to fix next.

If you are working out how to approach this, talk to our TRAFEED team.

References and primary sources

Footnotes

1. Expert Panel on Ensuring Research Security and Research Integrity, "Procedures Manual for Ensuring Research Security," December 2025 (Cabinet Office) https://www8.cao.go.jp/cstp/kokusaiteki/integrity/yushikisha/guidelines_v1.pdf ↩ ↩² ↩³ ↩⁴

2. MEXT, Science, Technology and Innovation Policy Bureau, "Direction of Concrete Measures on Research Security in MEXT-Related Programmes for Universities and Other Institutions," 18 December 2024 https://www.mext.go.jp/content/20250423-mxt_kagkoku-000039402_1.pdf ↩ ↩² ↩³

3. MEXT, Council for Science and Technology, Subdivision on Technology and Research Infrastructure, Committee on Promotion of Industry-Academia-Government Collaboration, "Report of the Conflict of Interest Working Group," November 2002 ↩ ↩² ↩³

4. Integrated Innovation Strategy Promotion Council, "Policy on Ensuring Research Integrity Against New Risks Accompanying the Internationalisation and Opening of Research Activities," 27 April 2021 https://www8.cao.go.jp/cstp/kokusaiteki/integrity/integrity_housin.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵

5. Inter-Ministerial Liaison Meeting on Competitive Research Funding, "Guidelines on the Proper Execution of Competitive Research Funding," 9 September 2005 (revised 17 December 2021) https://www8.cao.go.jp/cstp/kokusaiteki/integrity/shishin.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

6. Cabinet Office, "Checklist for New Risks Accompanying the Internationalisation and Opening of Research (Model)" (for researchers), 17 December 2021 https://www8.cao.go.jp/cstp/kokusaiteki/integrity/checklist1.pdf ↩ ↩² ↩³

7. Japan Science and Technology Agency (JST), "How to Apply for CREST, PRESTO and ACT-X FY2026 via the Cross-Ministerial R&D Management System (e-Rad)" https://www.jst.go.jp/kisoken/boshuu/teian/koubo/2026e-rad.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵

8. FY2024 MEXT Commissioned Project (University of Tokyo), "Collection of Near-Miss Cases in Research Integrity," March 2025 https://www.mext.go.jp/content/20250331-mxt_kagkoku-000019002_1.pdf ↩ ↩² ↩³

9. Japan Society for the Promotion of Science, "FY2026 Grants-in-Aid for Scientific Research Application Guidelines (Kiban A/B/C, Challenging Research, Early-Career Scientists)," 14 July 2025 https://www.jsps.go.jp/file/storage/kaken_kiban_2025_g_3687/r8_7_kobo.pdf ↩

10. Cabinet Office, "Policy on Ensuring Research Integrity (Overview)," September 2022 https://www8.cao.go.jp/cstp/kokusaiteki/integrity/gaiyo_202209.pdf ↩ ↩²

11. Cabinet Office, "Checklist for New Risks Accompanying the Internationalisation and Opening of Research (Model)" (for universities and research institutions), revised 29 June 2023 https://www8.cao.go.jp/cstp/kokusaiteki/integrity/checklist2r.pdf ↩ ↩²

12. Cabinet Office, "Policy on Ensuring Research Integrity: Summary of FY2025 Follow-Up Survey Results," December 2025 https://www8.cao.go.jp/cstp/kokusaiteki/integrity/ri_follow-up_fy2025/ri_fu_fy2025_sum.pdf ↩