OECD Business and Finance Outlook 2021 AI in Business and Finance

Foreword
Editorial
Abbreviations and acronyms
Executive summary
1 Trends and policy frameworks for AI in finance
	1.1. Introduction to AI in finance
	1.2. Insights from OECD.AI on AI diffusion in the financial sector
	1.3. Framing policy discussions on AI in finance
		1.3.1. Defining AI, its different types and its lifecycle
		1.3.2. Policy through the lens of the OECD AI Principles
		1.3.3. Policy through the lens of the AI system lifecycle
		1.3.4. Policy through the lens of the OECD Framework for the classification of AI systems
	1.4. National policies to seize opportunities and mitigate risks of AI in the financial sector
		1.4.1. Several national AI policies promote AI development and deployment in the finance sector
		1.4.2. Regulators are promoting safe and secure innovation while addressing specific challenges raised by the deployment of AI systems in financial services
	References
	Notes
2 AI in finance
	2.1. Introduction
	2.2. AI and financial activity use-cases
		2.2.1. Asset management  and the buy-side
		2.2.2. Algorithmic Trading
			AI algorithms, HFT and potential unintended consequences
		2.2.3. Credit intermediation and assessment of creditworthiness
			Risks of bias and disparate impact in credit outcomes
			Safeguarding mechanisms to mitigate risks of disparate treatment and bias
		2.2.4. AI in blockchain -based financial services
			Using AI to augment the capabilities of smart contracts
	2.3. Emerging risks and challenges from the deployment of AI in finance
		2.3.1. Data management, privacy/confidentiality and concentration risks
		2.3.2. Algorithmic bias and discrimination in AI
		2.3.3. The explainability conundrum
			Auditability and disclosure of AI techniques used by financial service providers
		2.3.4. Training, validation and testing of AI models to promote their robustness and resilience
		2.3.5. Governance of AI systems and accountability
		2.3.6. Other sources of risks in AI use-cases in finance: regulatory considerations, employment and skills
			Employment and skills
	2.4. Policy considerations
	References
	Notes
3 Human rights due diligence through responsible AI
	3.1. Introduction
	3.2. Overview of human rights impacts of AI
		3.2.1. Right to privacy
		3.2.2. Right to non-discrimination
		3.2.3. Right to fair trial and due process
		3.2.4. Freedom of expression
		3.2.5. Freedom of Association
	3.3. RBC applied to AI supply chain actors
		3.3.1. Six Step OECD Due Diligence Framework
		3.3.2. Roles/responsibilities of different supply chain actors
			Developers: Including key actors involved in data collection & processing, planning & design, model building & interpretation
			Vendors
			End Users
		3.3.3. Risk prevention/mitigation at different stages of the AI lifecycle
	3.4. National / International / Industry-led efforts to address AI risks
		3.4.1. Leveraging existing legislation
			Dual-use export controls
			Data protection
			RBC legislation
		3.4.2. AI-specific initiatives
	3.5. AI uses to support RBC
	3.6. Looking forward
	References
	Notes
4 Competition and AI
	4.1. Introduction
	4.2. Competition problems associated with AI
		4.2.1. AI and collusion
			AI and explicit collusion
			AI and tacit collusion
		4.2.2. AI and abuses of dominance
			AI developing or implementing anticompetitive strategies
			Anticompetitive design of consumer-facing AI
			Personalised pricing
		4.2.3. AI and mergers
	4.3. Challenges for competition policy in addressing AI-related competition problems
		4.3.1. Legal challenges
		4.3.2. Investigative challenges
		4.3.3. Competition policy approaches to addressing competition issues raised by AI
			Market studies and advocacy
			Co-operation with other regulators and stakeholders
		4.3.4. Considering reforms to current enforcement frameworks and new regulatory measures
	4.4. Conclusion
	References
	Notes
5 The use of SupTech to enhance market supervision and integrity
	5.1. Introduction
	5.2. Drivers and typology of SupTech developments
	5.3. The benefits of SupTech
		5.3.1. Improving detection capabilities
			Use cases by financial and securities regulators: better detecting market manipulation and insider trading
			Use cases by agencies involved in combatting corruption: better detecting criminal allegations and fraud
			Use cases by competition authorities: better detecting cartels and other types of anti-competitive practices
				Cartel screening
				Adapting techniques to investigate harm facilitated by algorithms
				Price monitoring tools
		5.3.2. Improving efficiency in enforcement actions
			Use cases by securities regulators: better determining compliance with disclosure requirements and guiding enforcement actions
			Use cases by agencies involved in combatting corruption and foreign bribery: better resolving cases
			Use cases by competition authorities: facilitating evidence review in cartel investigations and enhancing the monitoring of remedies
		5.3.3. Improving data collection
			Use cases by financial and securities regulators: improving regulatory reporting
			Use cases by competition and anti-corruption authorities: improving the collection of evidence during unannounced inspections
		5.3.4. Improving data management
	5.4. Challenges and risks of SupTech
		5.4.1. Data quality, standardisation and completeness
		5.4.2. Legal and procedural challenges
			Due process rights of companies as a legal challenge
			Data location as a legal challlenge
		5.4.3. Algorithmic models and human oversight
		5.4.4. Third-party dependencies, digital security and privacy concerns
		5.4.5. Legacy systems
		5.4.6. Financial and human resources, procurement rules, and barriers to change
	5.5. Considerations for devising adequate SupTech strategies
		5.5.1. Leadership, budget and skills
		5.5.2. Collaboration between authorities, regulated entities and technology service providers within and across jurisdictions
	References
	Notes
6 Managing access to AI advances to safeguard countries’ essential security interests
	6.1. Managing risk without stifling opportunities: new challenges require new solutions
	6.2. Managing essential security interests related to foreign acquisitions of AI assets in context
		6.2.1. New vulnerabilities emerge from international investment in advanced technology
		6.2.2. Greater scrutiny has not ended the international investment boom in AI
	6.3. Foreign investment in research: a new challenge calling for an adequate solution
	6.4. Managing the implied risks of openness without forgoing benefits
	References
	Notes