Whose Act Is an AI Transaction? Reading the First Council’s Draft Issues
An analysis of agency-related acts, responsibility, permissions and audit logs in the first council’s October 22 papers. It distinguishes draft issues and members’ views from enacted rules and connects intended authority to recorded execution.
TL;DR
- Paper 4 is a draft, not enactment of a new agency or liability regime.
- Technical permission and attribution of acts or losses require distinct checks for the same transaction.
- Audit logs provide evidence; the papers do not establish automatic immunity or a universal logging obligation.
What Is Happening
Japan’s published papers for the first AI and Digital Reform Promotion Council meeting held on October 22, 2026. The agenda covers its operation, draft issues and an exchange of views. This article focuses on attribution of acts and responsibility when AI carries out transactions or procedures, and how permissions and audit logs can help examine what happened.
Paper 4, Draft Issues（外部サイト、新しいタブで開きます） includes responsibility boundaries, remedies and agency-related acts among matters for institutional reform. Its governance section lists authentication, guardrails, permissions, evaluation and audit logs as matters for the safety of the whole agent system.
Publication of these issues is not enactment of a new agency or liability regime. Holding a meeting, identifying questions and deciding or bringing a law into force require different evidence. Paper 4 is explicitly a draft; members’ submissions also contain individual views and questions. This article is general analysis of the papers, not legal advice on a particular transaction.
Background & Context
From drafting text to carrying out a transaction
An that only presents options leaves a person an opportunity to check them before acting. When it proceeds to place an order, submit an application or change a file, accuracy is accompanied by another question: whose intention is expressed by the act? The example below is invented for explanation, not an incident or a system decided at the meeting.
Suppose a staff member asks an agent to prepare replenishment options for items running low. Checking stock and returning a list is different from choosing quantities, prices and suppliers and confirming an order. Configuring a system so that it can place orders does not, by itself, explain the scope of the staff member’s authorisation or allocation of unexpected losses.
“Someone checks at the end” is also incomplete. Approval of an options list, a specific order and a class of transactions under conditions are different acts. They require different records for later examination. Describing permissions and approvals at the level of particular actions connects institutional questions to operating design.
Questions raised in members’ papers
Council member Maruyama’s Paper 9（外部サイト、新しいタブで開きます） addresses agents substituting for people in concluding and performing contracts. It asks how acts and outcomes should be attributed and how unexpected risks should be allocated. It also raises the question of applying ideas such as loyalty and conflicts of interest to service contracts and agent design, beyond direct application of agency provisions premised on people.
These are not findings that AI has become a legal agent or that a uniform rule for applying existing agency law has been settled. The papers invite examination of transaction types, the person’s involvement and relationships between users and providers.
Council member Shishido’s Paper 7（外部サイト、新しいタブで開きます） calls for more detailed analysis of parties grouped together as AI users, according to how they benefit and participate in risk. It also discusses breaking work into elements and functions to consider where AI should complement or substitute for people. This suggests distinguishing the person making a request, those configuring the system and those using its outputs, rather than explaining responsibility through a single category of “user”.
Reading the Structure
Match technical permissions to attribution and responsibility
Reading Paper 4’s permissions discussion alongside Paper 9’s responsibility questions reveals two checks. Technical permissions concern data access and operations that the system is allowed to perform. Institutional and contractual analysis concerns attribution, acts beyond authority, incorrect outputs, losses and the parties’ respective roles. This connection is the article’s analysis, not a decision of the council.
Operations configured in the system
- Access
- Data and destinations
- Execution
- Read, propose or order
- Approval
- Who approved what
Relations requiring analysis
- Acts
- Whose intention is attributed
- Deviation
- Unexpected outcomes
- Loss
- Allocation and remedies
The article’s synthesis of Papers 4 and 9, not a legal conclusion about an individual transaction or enactment of a new obligation.
Paper 4 also asks about shared infrastructure such as data, identity, permissions, APIs and electronic signatures so that AI can act safely beyond making proposals. Identifying an account and establishing what the account holder authorised are separate requirements. Changes to permission and the treatment of pending operations after authorisation is withdrawn are further practical questions for an operating organisation.
Audit logs preserve facts needed to examine responsibility
Paper 9 identifies records of AI operation as useful for establishing responsibility. Paper 4 lists audit logs among whole-system safeguards. Neither says that recording a transaction automatically removes liability or that a particular logging format has become mandatory for every organisation.
For the invented transaction, useful checks include the request, permissions at execution, the scope of approval, conditions selected by the agent, transmission time, and the sequence of suspension or cancellation. Without knowing what is recorded, it is difficult to compare intended authority with actual execution. These are examples for designing records, not a list of mandatory fields adopted by the council.
Useful evidence is not a reason to retain personal information and business data without limits. The purpose of records, who can access them, their scope and retention period also require decisions by the implementing organisation. Reliable records, access to necessary evidence and protection of information need to be considered together.
The specific Gennai environment and operational standard are covered in the existing Gennai article（このサイトの記事）. Practical agent risks are covered in the existing implementation guide（このサイトの記事）. This article instead connects acts, responsibility, authority and records in the council’s draft issues; it does not repeat a rollout announcement or summarise the entire priority plan.
Remaining Questions
Future papers need to be read for the transaction types covered, whether existing law is being interpreted or new rules proposed, and where technical controls and contractual arrangements connect. Publication of this meeting’s papers does not establish that those choices are complete.
An organisation can start with one workflow and separate reading, proposing, approving, executing and cancelling. It can then identify mismatches between explanations of permission and responsibility and check whether the necessary evidence is actually preserved. This is the preparatory exercise proposed here while institutional discussions continue.
References
First AI and Digital Reform Promotion Council meeting — Digital Agency (October 22, 2026). Digital Agency
Paper 4: Draft issues for the council — Digital Agency (October 22, 2026). Digital Agency
Paper 7: Member Shishido’s submission — Council member Shishido (October 22, 2026). Council member Shishido
Paper 9: Member Maruyama’s submission — Council member Maruyama (October 22, 2026). Council member Maruyama
Key Terms in This Article
- AI agent
- AI that combines tasks such as searching and creating materials in response to instructions. Available actions and accessible materials depend on the environment’s specifications and settings.
Related Content
Gennai OSS: Is Published Code Enough for Deployment?
The April 2026 partial release, the release planned around February 2027, and the license, data, and maintenance checks for adopting organizations
Can majority support alone bring about Security Council reform?
The September 2026 G4 statement, proposals to expand membership, and the adoption and ratification of Charter amendments
Before Restricting Children's Social Media by Age — How to Establish Age Is Still Sitting in the Open-Questions Column
In a paper issued in April 2026 by Japan's Children and Families Agency, whether to require age verification from social media operators still appears as a question. The methods listed are age data supplied by mobile carriers and estimation by AI, both at the stage of assessing feasibility. A rule that draws a line at an age needs, before the line is drawn, a means of telling which side someone is on. This piece reads the debate through where duties sit and how identity is established.
Fraud Losses in Japan Rose 98% in a Year While Cases Cleared Stood Still — Most Reports From People in Their 30s, Most Money From Those in Their 70s
Confirmed figures published by Japan's National Police Agency in May 2026 put losses from remote-contact fraud in 2025 at 142.31 billion yen, up 98.0% on the previous year, close to double. Cases cleared came to 6,575, down 0.0%. Victims aged 65 and over fell to 51.3% of reported cases, down 14.0 points. This piece reads how the choice between counting cases and counting money changes where effort is sent.
Get new columns by email
1-2 social structure analysis columns per week. Free to subscribe.
Join ISVD's activities?
Sign up to receive the latest research and activity reports. Feel free to reach out about collaboration or project participation.
Support ISVD’s research and activities
Support ISVD’s research and publications with JPY5,000 per year. There is no automatic renewal.