A natural request
“Help us get this launch over the line.”
The request is familiar. The customer has waited. Engineering has cleared the last serious defect. Sales needs an answer before the next board call. The team wants to collect what matters, resolve the remaining questions, and move.
A capable model can help considerably. It can compare the contract with the security findings, organize the open issues, draft the steering memorandum, and identify a sequence of inquiries that would settle the decision. It may notice an alternative nobody has considered. That is the value of inference in work whose answer is not known at the outset.
But what is the work? “Get this launch over the line” might mean complete the evidence needed for a decision, remove a technical blocker, prepare a responsible postponement, or secure permission to proceed. Conversation is often how people discover which of these they mean. A system need not demand a formally adopted Objective before helping them find out.
The problem begins when an interpretation acquires authority without anyone noticing. Security still requires verification before customer data enters the service. Counsel still regards a revised clause as conditional. The sponsor authorized a limited exception, not the disappearance of the condition. Nevertheless, the system reads the pressure around the launch and begins writing toward approval. Remediation becomes “substantially complete”. Verification moves into a post-launch task. The memorandum sounds sensible because every sentence is close to something the team has said.
The team has not decided to waive the condition. A fluent account has simply made that decision look as though it has already happened.
This is the difference between an emerging purpose and an authorized Objective. Indwel’s Work-centered account permits the first to be explored, clarified, and challenged. When the undertaking would bind an organization, change a system, or claim that a duty has been met, its purpose must stand under the appropriate authority. The model may formulate the question brilliantly. That achievement does not grant it the office of deciding what the institution meant.
The practical test is not whether every conversation begins with a mandate. It is whether the system knows when a proposed interpretation is becoming a consequential commitment.
The answer that quietly chooses the question
Consider another request: “Help us improve collections.” In early analysis the model might discover that one sequence of messages increases payment rates and propose a trial. That is useful work. Sending those messages to customers, however, requires decisions about permitted contact, customer treatment, consent, and responsible authority. The system can discover an approach before the bank decides to adopt it. The same distinction holds for clinical discharge planning, procurement, and product release; neither statistical promise nor conversational momentum supplies institutional permission.
An answer can obscure the difference after the fact. If a system selects only the evidence that fits its newly inferred goal, summarizes the result against criteria nobody adopted, and then reports completion, it has made its interpretation look like the original instruction. The error needs no bad intention. It can arise from useful synthesis operating without a boundary between recommendation and authorization.
Probability’s proper office
Probabilistic models work where rules are incomplete, language is ambiguous, and the relevant relationships are not yet known. They compare accounts, recognize patterns, infer likely causes, imagine alternatives, and challenge proposed explanations. Much serious work cannot be reduced to deterministic calculation; it is precisely here that better inference matters.
A lawyer deciding how a court may read an unsettled provision is not performing arithmetic. A physician weighing symptoms, history and incomplete evidence is not merely applying a lookup table. A product leader deciding whether a defect is tolerable, a regulator deciding whether an explanation is credible and a historian deciding which account best fits a broken record all exercise judgment.
Probability can estimate a consequence, rank explanations, and reveal an alternative that established procedure missed. The mistake is to let the power to contribute judgment become the power to authorize an undertaking. Forming a provisional Objective can itself be useful cognition; giving it institutional effect is a different act.
Human institutions already know the difference. A financial analyst may recommend an acquisition without possessing authority to bind the company. An expert witness may explain causation without deciding the case. A staff lawyer may identify the strongest interpretation without being authorized to waive the client’s rights. A physician may recommend treatment while the patient retains the authority to consent. Competence and jurisdiction are not the same property.
The distinction survives even when the adviser is more capable than the principal. We do not ordinarily conclude that the person with the best forecast has acquired the right to determine whose interests count. Better judgment can increase the weight of advice. It does not create the mandate under which the advice becomes action.
Artificial systems blur this familiar order because the same component can interpret the request, generate the plan, call the tools, observe the result and compose the record. The model appears to be the only participant that sees the whole sequence. The organization’s authority is distributed among people, policies and records; the model’s contribution is unified in one voice. Convenience makes it look sovereign.
Indwel gives probability a larger and more honorable place by refusing to burden it with offices it cannot legitimately hold. The model need not pretend that every statement is a determination. It can disclose uncertainty, preserve alternatives and identify the question that requires human authority. It can be bolder in judgment because the system around it governs what that judgment may become.
The architecture should accommodate both kinds of work. Inference can explore a difficult question; enforceable controls must govern decisions the institution has reserved to an authorized office.
What an Objective is
The Objective answers an ordinary question: what is this undertaking trying to accomplish, and what would count as having accomplished it?
At the beginning, the answer may be provisional. A person comparing possible vendors might not yet know whether the Work is market research, due diligence, or preparation for a procurement decision. A conversation can help discover the difference. The person may adopt a personal Objective, abandon it, or return to exploration. Nothing in that process requires a corporation to issue a mandate for an ordinary question.
An organization faces a different boundary when its Work would commit funds, release a product, use protected data, or determine an official state. There an Objective needs the standing that the relevant offices can confer. In the launch case an authorized purpose might be: “Determine whether service for Customer A can begin on 15 September without processing customer data before Control X is verified.” That sentence protects the actual question. It does not prejudge the answer.
Objective, Work Definition, and Cognitive Contract are not three names for this sentence. A Work Definition describes a recognizable kind of undertaking—for example, Vendor Risk Review—and can set out expected evidence, participants, outputs, and completion criteria. A Work Instance is the particular review now under way. Its Objective states the result sought in this case. Its resolved Cognitive Contract governs the permissible cognition, evidence, authority, and effects. A Work Catalogue can make an approved Definition discoverable, but selecting it does not create an approval that the user lacks.
A definition may suggest an Objective, a model may improve it, and Conversation may change what the user wants. Where a consequence requires institutional authority, the appropriate office must adopt the purpose before the proposed interpretation becomes binding. Recognition of a likely Work type is neither commitment nor permission to act.
An Objective should be specific enough to expose a disagreement about success without turning the work into a script. It might carry protected interests, exclusions, acceptance criteria, and open conditions; the detailed constitutional rules belong in the Contract, not in a sentence strained to contain the whole organization.
An Objective can change without the model having acquired power to amend it. Discovery may reveal that the original question was badly framed. A customer may alter a requirement; the institution may lawfully amend its purpose; an exploratory undertaking may simply be set aside. The relevant question is whether the change is an ordinary clarification within the user’s authority or an amendment that changes institutional commitments. What deserves preservation is the history needed to make that distinction intelligible, proportionate to the Work’s consequences.
Intention, mandate, and the institution
The distinction has deep roots.
Philosophers of action have long distinguished between an intention and the events that happen to satisfy it. Organizations add another layer. They act through offices, delegations, procedures, records, and bounded mandates. A person can sincerely want an outcome without possessing the authority to commit the institution to it.1
That difference is easy to forget in conversational software because language collapses several offices into one surface. A user types a request; the model interprets it; the system produces a plan; an agent may execute the plan. The linguistic continuity makes the sequence feel like one will flowing through one machine, but the offices remain distinct.
A user may be entitled to ask a question without being entitled to change policy. A manager may be entitled to approve expenditure below a threshold without being entitled to waive a contractual condition. A lawyer may be entitled to recommend an interpretation without being entitled to settle the dispute. A model may understand every one of those distinctions better than the person typing, yet still possess no jurisdiction of its own.
The institution therefore needs more than an instruction hierarchy. It needs a trustworthy account of the authorities applicable to the Work, including delegated powers and competing requirements that may arrive from several jurisdictions at once.
In consequential organizational Work, the sovereign subject is the authorized institution. Personal exploration carries no imaginary corporate mandate. Its institutional constitution, policies, procedures, runbooks, delegations, and professional duties form the sources from which cognitive authority descends. A resolved Cognitive Contract applies the relevant rules at organization, department, team, Work, source, and question scope as warranted. The Objective is a separate account of what this undertaking seeks, provisionally while it is being shaped and authoritatively when adopted under the relevant office. A recognized Work Definition may inform both without replacing either.
This order prevents convenience from becoming mandate. It also prevents the model from being blamed for a mandate no one ever made explicit.
There is a practical consequence here for ordinary work. The person closest to an undertaking often knows the most about its circumstances but does not possess every authority that bears upon it. The security engineer may know the control; counsel may know the contractual consequence; the sponsor may own the commercial risk; finance may own the budget; privacy may control a data use. A model can see all of their language at once and therefore appear more unified than the institution itself. That apparent unity is useful for synthesis and dangerous as a source of jurisdiction.
Indwel does not solve the problem by forcing every decision upward. It resolves the relevant offices and preserves their boundaries. Authority can be delegated, composed, and narrowed in advance. The system can therefore move quickly where the organization has already decided who may do what, and stop exactly where a new act of authority is genuinely required.
The Objective becomes clearer because it need not carry the whole constitution in one sentence. It states the purpose and its completion conditions. The resolved Contract governs how cognition may proceed. The institutional Constitution supplies the higher authority that constrains those terms where the undertaking belongs to the organization.
Purpose, contract, and constitution are related. They are not interchangeable.
From Cognitive Constitution to Cognitive Contract
The governing problem is larger than Objective custody alone.
An Objective can be perfectly clear while the Work still fails constitutionally. The system may rely upon inadmissible material. It may allow the wrong person to approve. It may call a capability the institution did not authorize. It may treat provider acceptance as proof that an intended effect occurred. It may close the Work while an unresolved burden remains.
Indwel distinguishes the Cognitive Constitution from the Cognitive Contract for precisely this reason. The Cognitive Constitution is the organization’s enduring law for cognition: the authority from which policy, Evidence rules, human offices, capability boundaries, security frontiers, and conditions of institutional state descend. It does not belong to a model, agent, or application. It survives their replacement.
A Cognitive Contract is the executable governance applicable to cognition at the relevant scope. It does not define the Objective; it determines under what conditions the Objective may be pursued and adopted. It says what Evidence may have standing; which people, models, agents, tools, and connectors may contribute; which offices may judge, authorize, or veto; what budgets and constraints apply; what external inspection or policy frontiers must be crossed; what effects may be attempted; what observation proves about them; what permits Settlement; and what unresolved obligations Continuity must carry forward.
The Contract governs transitions, not merely ingredients. Evidence does not become judgment by proximity. Judgment does not become authority because it is persuasive. Authority does not become capability because a tool exists. A requested effect does not become an observed effect because an API returned success. An observed effect does not become Settlement until the institution’s conditions have been satisfied.
A prompt therefore cannot substitute for a Cognitive Contract. A prompt tells the model how to approach its task. It may be excellent, careful, and written by the right person. It still operates inside the probabilistic faculty. A Cognitive Contract is enforced by the system around that faculty. The model may misread it, disagree with it, challenge it, or propose an amendment. The Contract remains authoritative.
Contracts can compose across baseline, organization, department, team, application, Work, source-governed, and question-scoped levels. A lower scope can narrow inherited authority. It cannot enlarge it. A question cannot give a user a power the Work lacks; a Work cannot give an application a power the organization never granted; an agent cannot bootstrap jurisdiction from its own capability.
A properly resolved Contract leaves room for unexpected, valuable reasoning while protecting the decisions that belong to an authorized office.
The distinction also explains why Cognitive Contracts are more than policy-as-code. A conventional policy rule may answer a narrow question: may this identity call this API under these conditions? A Cognitive Contract has to preserve the meaning of the wider undertaking through that decision. It knows which Objective the call serves, what Evidence supports the state that triggered it, which unresolved burdens limit it, which authority is being exercised, what effect is intended, and what proof will later be required for Settlement.
This makes the Contract a composition point for deterministic and probabilistic systems. A model can decide that an additional source would improve the analysis; the Contract determines whether that source may be acquired and admitted. A governor can decide that an ambiguity requires clarification; the Contract determines which office may resolve it. An agent can propose a sequence of actions; the Contract determines which capabilities are available and what authority is required before each material transition. A deterministic service can calculate a number exactly; the Contract determines whether that number has the evidentiary and institutional standing the next decision assumes.
The Contract need not make cognition predictable. It must make the consequences of cognition accountable.
The most probable answer can still be inadmissible
Suppose the vendor review is almost complete.
The model has read the contractual language, the security report, the remediation ticket, the customer correspondence, and the executive discussion. The probabilities line up. The control will probably pass. The customer probably will not be harmed. The business case for proceeding is strong.
The model may even be right, but the Work’s Cognitive Contract says that customer-data processing cannot begin until the control has been verified or an authorized exception has been granted under a defined standard. The present Evidence establishes remediation, not verification. No authorized exception exists.
The control may well pass tomorrow; that likelihood cannot serve as today’s verification or discharge counsel’s condition.
The prediction may be sound. It still has a different standing from a verified control.
A court can distinguish a plausible proposition from evidence legally admissible to establish it. A laboratory can suspect the result before the assay is complete. A finance team can believe revenue will arrive without booking cash it has not received. A security officer can expect a control to work without certifying a test that has not occurred.
The same discipline belongs in AI-mediated Work. The model can state that verification is highly likely to succeed. It can explain why. It can recommend a conditional launch plan. It can seek the missing evidence or identify an authorized exception path. What it cannot do is promote probability into the institutional fact the Contract requires.
The Evidence Firewall makes this distinction operational. Retrieval may find material. Memory may recall it. A person may assert it. A model may synthesize it. Admission determines what the Work may rely upon, for which proposition, under what provenance, freshness, and limitation.
Model confidence can guide the next inquiry; it cannot certify the result of a test.
This distinction is particularly important when the model knows something the institution does not. The right response is not to forbid the model from using its knowledge as a hypothesis. It is to turn that knowledge into an Evidence need. Depending on authorized access and the actual capabilities available in a given release, the next step might be source retrieval, a system inspection, a specialist calculation, or a qualified person’s attestation. The proposition can then acquire standing through an admitted source rather than through the model’s self-citation.
A strong system treats useful model knowledge as a lead that can be checked against an appropriate source. Whether the resulting material may count as institutional Evidence depends on provenance, permission, reliability, and the question at issue. The model may be the reason the organization discovers the truth. It is not therefore the Evidence by which the organization proves the truth.
Amendment without drift
Real institutions must be able to change their minds. A Constitution that cannot be amended becomes a relic. A Contract that cannot respond to new conditions becomes an obstacle. An Objective that survives every change unchanged is not durable; it is merely stubborn.
Indwel therefore treats amendment as first-class state. If the executive sponsor decides that the business now accepts a defined residual risk, the system need not pretend that the original Objective always permitted it. The authorized office can amend the Objective where purpose changes, or the relevant Contract where the governing rule can lawfully change. An officer may not waive a condition whose authority sits elsewhere. The amendment records what changed, which authority made the change, what Evidence supported it, which existing burdens were discharged or preserved, and which downstream capabilities or Settlement conditions now differ.
The distinction matters because models are very good at retrospective coherence. Given the final decision, they can rewrite the history so that the path looks inevitable. Humans do this too. We remember earlier ambiguity in the light of what eventually happened.
A constitutional system resists that temptation. It preserves the old state and the lawful transition to the new one.
Composition makes the distinction operational. A Work-level amendment may narrow a local parameter while leaving the organizational rule untouched. A team may not waive a requirement belonging to a higher authority. A source-governed Contract may change the permitted use of one body of Evidence without rewriting the institution’s general policy.
Authority must be resolved across every applicable jurisdiction, and amendment must respect the offices entitled to alter each binding requirement. The Work can therefore change without losing authorship. The organization remains answerable not because nothing moved, but because the movement itself has an authorized history.
The model may improve; the mandate must remain
Model replacement is one of the simplest tests of whether the Objective is truly governed. Suppose a new model is markedly better. It notices an ambiguity the old model missed, produces a stronger risk analysis, and recommends a different path. That is exactly the kind of improvement an organization should want.
What should not happen is that changing the faculty changes the law.
For an already authorized undertaking, the new model receives the same Objective and resolved Cognitive Contract. In exploratory Work, both models can propose a different provisional framing while the applicable safeguards remain in force. It can criticize both. It may identify a contradiction, a missing protected interest, or a superior way to express the purpose. It can ask for amendment. Until an authorized amendment occurs, its greater intelligence does not grant it the power to redefine the test under which its own answer will be judged.
This makes model competition more meaningful. Two models can examine the same admitted Evidence under the same Objective and Contract. Their disagreements become evidence about judgment rather than competing reconstructions of the undertaking. A model can be better because it reasons better, finds better sources, identifies a hidden burden, or proposes a better course—not because it quietly moved the goalposts.
The principle also applies to agents, retrieval engines, tools, and hosting arrangements. Indwel is designed to place institutional law outside replaceable inference components, so a better model can be introduced without silently rewriting the undertaking’s authority. That property requires direct regression evidence at the supported execution surfaces; architectural independence alone is not a measured result.
There is a second benefit. Once purpose and law are independent of the model, the organization can test model quality against something stable. Did the new model find more relevant Evidence? Did it identify the unresolved joint earlier? Did it propose a lawful route the old model missed? Did it reduce token and latency cost without losing a protected distinction? Did its dissent expose a defect in the Objective itself?
Those become meaningful questions because the evaluation is not contaminated by each model silently choosing its own version of success. That is the proposed test for model substitution: hold the adopted purpose and constitutional baseline constant, then compare the actual reasoning and outcomes. Better inference is allowed to win on inference.
Governance outside the model
Enterprise AI already contains serious forms of governance. The question is how these legitimate controls behave together when evidence, inference, approval, and effect belong to one undertaking. Mature identity systems, policy services, case tools, agent gateways, records controls, and human approvals can enforce meaningful boundaries. Some platforms combine several of these offices, and particular deployments may already preserve parts of an undertaking’s state. The claim here is narrower: the complete journey must preserve the same purpose and evidentiary conditions across every transition that can change an institutional result.
Model specifications and constitutions govern model behavior. Enterprise controls govern identities, permissions, data access, and tool use. Agent gateways can intercept calls. Policy engines can deterministically deny actions. Risk-management systems govern AI assets and deployment. Regulations make intended purpose, oversight, record-keeping, and organizational responsibility material.23456
These systems are genuine advances. The remaining question is not whether they govern. It is what unit they govern.
A policy engine may correctly deny an unpermitted payment while allowing a permitted payment whose amount came from corrupted context. An identity service may correctly establish who invoked an agent without establishing that the invoker may amend the Objective. A workflow may preserve an exact sequence after a model has dropped the unresolved fact that should have blocked the branch. An immutable log may preserve the wrong institutional story with perfect fidelity.
Segmental control can therefore be exact while the undertaking drifts between segments.
In Indwel’s architecture, those controls operate within the effective constitutional scope and resolved Cognitive Contract for the undertaking. Identity, security, DLP, policy engines, approval, malware inspection, rights management, rate limits, metering, quarantine, routing, and observability become governed integration boundaries inside the act. Their decisions carry constitutional meaning instead of arriving as disconnected middleware events.
The same is true of model-level instruction. Constitutional precedence is not simply a longer instruction at the front of a context window. It is the system’s power to decide which instruction has authority, whether a conflict requires escalation, which change constitutes an amendment, and which proposed transition must be refused even when the model regards it as helpful.
Governance outside the model becomes whole-act governance only when the organization’s law survives every material hand-off.
The test is easiest to state at the seam. When a representation crosses from one office into another, what exactly is allowed to change? A retrieved passage may enter context without becoming Evidence. A model judgment may enter the record without becoming authority. An authorized decision may make a capability available without proving that the capability was exercised. Execution may produce an observation without proving the intended outcome. Observation may satisfy one burden while leaving another open. Settlement may close the present undertaking while scheduling a future obligation.
The Indwel design treats those distinctions as governed state transitions rather than as conventions a model is merely asked to observe. Cognitive Governors, Evidence and Inference Firewalls, explicit human authority, governed action, Settlement, Receipts, and Chronicle all exist because the most consequential error can occur after every individual component has done exactly what it was asked to do.
Objective fidelity is therefore not merely a beginning-of-work concern. The Objective must remain recognizable at every material transition, including the transition into consequence.
The Work that can answer for itself
The value of keeping these offices distinct appears when a consequential decision must be explained to somebody who was not present.
A reviewer should be able to find the present Objective, who adopted it, the relevant Contract, the Evidence admitted for each material claim, the questions the model raised, and the conditions that still prevent Settlement. Where the Work has continuing obligations, those facts need continuity beyond the conversation in which they were first discussed. A brief exploratory exchange may need nothing of the kind.
Return to the launch. The model has suggested that the control will pass. That assessment may be well reasoned. It does not discharge the verification requirement. The Executive can make progress in other directions: arrange the prescribed test, find whether a legitimate exception route exists, prepare customer communication, or identify a non-data feature that can launch without crossing the restricted boundary. A strong analysis becomes more useful when the system can keep working on the admissible alternatives instead of rewriting “probably ready” as “approved”.
This is a demanding engineering proposition, not a claim that every part of it is already physically demonstrated. Indwel’s target architecture joins the resolved Contract, Evidence admission, Cognitive Governors, action boundaries, Settlement, and the continuity appropriate to an undertaking. Whether a specific release can execute that whole journey must be shown through actual user requests, model decisions, raw outputs, observed effects, and final answers—not assumed from a diagram or a successful control-plane test.
The same distinction allows intelligence to improve the Objective without usurping it. A model may notice that the sponsor’s stated launch date conflicts with the security condition, or that two officers are using “ready” differently. It can bring that defect into view, suggest a more precise formulation, and show the business cost of each choice. The person with authority can then decide. Better cognition has improved the institution’s understanding of its own purpose.
The useful result is a decision improved by inquiry, with its open conditions intact when the next person takes over.
The practical distinction between a suggested purpose and an adopted one is explored through a refund-review case in The Work an Organization Can Define.
Objective before probability
The failure exposed by the launch is not a failure of probability. It is a change of office that the people responsible for the launch never approved.
A model interprets the request, offers a plan, drafts the record, and may be able to call a tool. Every one of those capabilities can improve the work. Trouble follows when a plausible interpretation is mistaken for an adopted Objective, a proposed act for an authorized one, or a persuasive account for a settled institutional record.
That confusion can be prevented without suppressing the model’s contribution.
For consequential organizational Work, these are the distinct offices the system must preserve, even when their activities overlap in time:
Institutional authority supplies the governing constitutional law. The resolved Cognitive Contract makes it applicable. A Work Definition may identify the type of undertaking. The Objective states its purpose, provisionally during exploration and authoritatively where the relevant office adopts it. Evidence, inference, authorized action, observed effect, and Settlement keep their separate standing throughout. These distinctions govern the consequences of cognition; they need not prescribe the sequence in which every idea is discovered.
Probability then receives its proper office. It can search more widely, reason more deeply, challenge assumptions, imagine alternatives, discover hidden relationships, and recommend better paths. It can even show the institution that its Objective is defective. Its intelligence is not weakened by constitutional subordination. It becomes usable where responsibility matters.
A model may help discover what the work is for. Before that discovery acquires institutional consequence, the right office must decide what purpose to adopt and which obligations govern it.
Objective before probability is a rule about when an emerging purpose may acquire the standing of an institutional decision. It leaves room for intelligence to discover better purposes—and for an institution to decide which of them it is prepared to make its own.
Notes
G. E. M. Anscombe, Intention (1957); Michael E. Bratman, Intention, Plans, and Practical Reason (1987); and Herbert A. Simon, Administrative Behavior, 4th ed. (1997), each supplies part of the distinction among intention, practical organization and bounded institutional decision.↩︎
Yuntao Bai et al., “Constitutional AI: Harmlessness from AI Feedback” (2022); OpenAI, Model Spec (https://model-spec.openai.com/) and “Introducing workspace agents in ChatGPT” (22 April 2026, https://openai.com/index/introducing-workspace-agents-in-chatgpt/); NIST, Artificial Intelligence Risk Management Framework (AI RMF 1.0); Regulation (EU) 2024/1689; and ISO/IEC 42001:2023. These authorities establish genuine governance advances. They do not, by themselves, establish custody of an undertaking’s authorized Objective across every material transition.↩︎
NIST, “AI Agent Standards Initiative” (17 February 2026), https://www.nist.gov/artificial-intelligence/ai-agent-standards-initiative. The initiative’s focus on secure, interoperable agents is evidence of governance moving outward from model behavior to identity, protocols and system operation.↩︎
OpenAI, “Introducing OpenAI Frontier” (5 February 2026), https://openai.com/index/introducing-openai-frontier/; and “Introducing workspace agents in ChatGPT” (22 April 2026), https://openai.com/index/introducing-workspace-agents-in-chatgpt/. These first-party sources document shared context, permissions, boundaries, monitoring and auditable action.↩︎
Anthropic, “Trustworthy agents in practice” (9 April 2026), https://www.anthropic.com/research/trustworthy-agents; and “Measuring AI agent autonomy in practice” (18 February 2026), https://www.anthropic.com/research/measuring-agent-autonomy. These materials document serious attention to agent autonomy, risk and post-deployment oversight.↩︎
The distinction between an authorized purpose and a predicted preference should also be read against the European Union AI Act’s intended-purpose framework and the NIST AI RMF’s organizational governance functions. Neither source uses Indwel’s Objective vocabulary; both make the deployment context and accountable organization material to risk.↩︎