Lifecycle alignment

Where each framework speaks across the AI lifecycle.

Some frameworks address framing and policy; some live inside the model; some are about what happens after deployment. This view places each item at the lifecycle stages it actually addresses, without stretching it to fit cells it does not cover.

Frameworks

Framework / Stage	Framing Problem definition, stakeholder mapping, go/no-go decision.	Data Sourcing, provenance, labeling, rights, preprocessing.	Model Training, evaluation, fairness/safety testing.	Deployment Integration, rollout, change management, disclosures.	Monitoring Ongoing performance, incident response, drift.	Retired Decommissioning, data disposition, archival.
UNESCO Adopted 2021, reaffirmed 2024	Awareness & literacyAwareness and literacy Environment & ecosystemsEnvironment and ecosystem flourishing Human oversight & determinationHuman oversight and determination Multi-stakeholder governanceMulti-stakeholder and adaptive governance Proportionality & do-no-harmProportionality and do no harm + 4 more	Environment & ecosystemsEnvironment and ecosystem flourishing Fairness & non-discriminationFairness and non-discrimination Privacy & data protectionRight to privacy and data protection Policy Area 3Data policy Policy Area 6Gender	Environment & ecosystemsEnvironment and ecosystem flourishing Fairness & non-discriminationFairness and non-discrimination Privacy & data protectionRight to privacy and data protection Safety & securitySafety and security Transparency & explainabilityTransparency and explainability + 2 more	Awareness & literacyAwareness and literacy Environment & ecosystemsEnvironment and ecosystem flourishing Fairness & non-discriminationFairness and non-discrimination Human oversight & determinationHuman oversight and determination Multi-stakeholder governanceMulti-stakeholder and adaptive governance + 7 more	Fairness & non-discriminationFairness and non-discrimination Human oversight & determinationHuman oversight and determination Multi-stakeholder governanceMulti-stakeholder and adaptive governance Responsibility & accountabilityResponsibility and accountability Safety & securitySafety and security + 2 more	Privacy & data protectionRight to privacy and data protection Responsibility & accountabilityResponsibility and accountability Policy Area 8Environment policy
OECD 2022 (informs 2019/2024 OECD AI Principles)	1.1 UsersUsers of the AI system 1.2 Affected stakeholdersAffected stakeholders 1.3 Human rights & democratic valuesHuman rights and democratic values 2.1 SectorIndustry sector 2.2 CriticalityCriticality of function + 2 more	3.1 ProvenanceData provenance 3.2 Dynamic vs. staticDynamic or static input Dimension 3Data & Input	4.1 Model typeModel type and family 4.2 ExplainabilityExplainability Dimension 4AI Model	1.1 UsersUsers of the AI system 1.3 Human rights & democratic valuesHuman rights and democratic values 4.2 ExplainabilityExplainability 5.1 AutonomyAutonomy of action 5.2 Combining tasksCombining tasks and actions + 2 more	1.2 Affected stakeholdersAffected stakeholders 3.2 Dynamic vs. staticDynamic or static input Dimension 1People & Planet	OECD does not directly address the retired stage in the indexed content.
IEEE CertifAIEd Program v2	A-2Defined roles and responsibilities AccountabilityAccountability	P-1Data minimization P-2Lawful basis and consent Algorithmic BiasAlgorithmic bias PrivacyPrivacy	B-1Bias measurement B-2Mitigation evidence Algorithmic BiasAlgorithmic bias PrivacyPrivacy TransparencyTransparency	A-1Recourse and redress A-2Defined roles and responsibilities P-2Lawful basis and consent T-1Stakeholder-tailored disclosure T-2Decision traceability + 4 more	A-1Recourse and redress B-1Bias measurement T-2Decision traceability AccountabilityAccountability Algorithmic BiasAlgorithmic bias + 1 more	AccountabilityAccountability PrivacyPrivacy
OCAP® Living principles	ControlControl OwnershipOwnership Stewardship 1Community engagement before scoping Stewardship 2Benefit and risk to community	AccessAccess ControlControl OwnershipOwnership PossessionPossession	ControlControl	AccessAccess ControlControl PossessionPossession	ControlControl Stewardship 2Benefit and risk to community	AccessAccess ControlControl PossessionPossession Stewardship 3Repatriation and disposition
MITRE ATLAS Pinned 2026.Q1	AML.M0020Disclosure discipline	AML.TA0002Reconnaissance AML.TA0003Resource Development AML.T0000Search victim-owned websites AML.T0006Acquire public ML artifacts AML.T0020Poison training data + 2 more	AML.TA0002Reconnaissance AML.TA0003Resource Development AML.TA0006Persistence AML.TA0010Exfiltration AML.T0000Search victim-owned websites + 7 more	AML.TA0002Reconnaissance AML.TA0004ML Model Access AML.TA0005Execution AML.TA0006Persistence AML.TA0007Defense Evasion + 19 more	AML.TA0007Defense Evasion AML.TA0011Impact AML.T0029Denial of ML service AML.M0018Content filtering	MITRE ATLAS does not directly address the retired stage in the indexed content.
NIST AI RMF 1.0 + Generative AI Profile (NIST AI 600-1)	GOVERNGovern MAPMap GOVERN 1Policies, processes, procedures GOVERN 2Accountability structures GOVERN 3Workforce diversity, equity, inclusion + 7 more	MAPMap	MAPMap MEASUREMeasure MAP 3Capabilities and limitations MEASURE 2Trustworthiness characteristics MEASURE 2.11Fairness and bias evaluated + 2 more	GOVERNGovern MANAGEManage MEASUREMeasure GOVERN 1Policies, processes, procedures GOVERN 2Accountability structures + 4 more	GOVERNGovern MANAGEManage MEASUREMeasure GOVERN 5Engagement with AI actors MANAGE 1Risk responses planned + 6 more	GOVERNGovern MANAGEManage MANAGE 4Continual monitoring & improvement

Empty cells are not failures of a framework — they are honest signals about scope. Pulling a framework into a stage it does not address is precisely the move Lattice refuses.

Where each framework speaks across the AI lifecycle.

Command palette