A Constitutional Quagmire - Ethical Minefields of AI, Cyber, and Privacy.pdf
0 likes181 views
A Constitutional Quagmire - Ethical Minefields of AI, Cyber, and Privacy
A Constitutional Quagmire - Ethical Minefields of AI, Cyber, and Privacy.pdf
- 1. SESSION ID: #RSAC Daniel Garrie A Constitutional Quagmire: Ethical Minefields of AI, Cyber, and Privacy CEO; Neutral; Adjunct Professor Law and Forensics; JAMS ADR; Harvard LAW-R01
- 2. #RSAC Disclaimer Presentations are intended for educational purposes only and do not replace independent professional judgment. Statements of fact and opinions expressed are those of the presenters individually and, unless expressly stated to the contrary, are not the opinion or position of RSA Conference or any other co-sponsors. RSA Conference does not endorse or approve, and assumes no responsibility for, the content, accuracy or completeness of the information presented. Attendees should note that sessions may be audio- or video-recorded and may be published in various media, including print, audio and video formats without further notice. The presentation template and any media capture are subject to copyright protection. © 2024 RSA Conference LLC or its affiliates. The RSA Conference logo and other trademarks are proprietary. All rights reserved. This is not legal advice, nor should it be considered legal advice. This presentation and the comments contained therein represent only the personal views of the participants, as spoken and do not reflect those of their employers or clients. 2
- 3. #RSAC Legal Concerns of AI Use: Scenarios To Consider
- 4. #RSAC Legal Concerns with AI The development and use of AI technologies raises a number of legal concerns: – Privacy and security – Transparency, explainability, and accountability – Intellectual property protections – Fairness and acknowledgment of bias – Inclusiveness – Reliability and safety 4
- 5. #RSAC Privacy Scenario A healthcare AI application designed to provide personalized treatment recommendations is found to be: – Collecting, – Storing, and – Processing patients' data without explicit consent. 5
- 6. #RSAC Actionable Insights to Address Privacy Challenges Implement "privacy by design" principles, ensuring data protection is a core element of AI development. Obtain explicit consent from users for data collection and processing, clearly explaining the purpose and use. Audit data handling practices to ensure compliance with data protection laws. Develop a robust data breach response plan, including timely notification to affected individuals. 6
- 7. #RSAC Cybersecurity Scenario An AI-based system is compromised due to a previously unknown vulnerability. Hackers bypass the AI's detection mechanisms and access sensitive customer data. 7
- 8. #RSAC Actionable Insights to Address Cybersecurity Challenges Establish a comprehensive cybersecurity framework for the AI systems, including regular security assessments. Develop and implement a rapid response plan for AI-related cybersecurity incidents. Update and patch AI systems against new threats. Ensure legal and regulatory compliance in data breach notifications and remediation efforts. 8
- 9. #RSAC Trade Secret Protection Scenario A leading tech company claims a competitor is using its proprietary algorithms to improve their own AI systems. 9
- 10. #RSAC Actionable Insights to Address Trade Secret Protection Challenges Implement stringent access controls and encryption for sensitive AI algorithms and data. Regularly review and update intellectual property protection strategies for AI technologies. Pursue legal remedies swiftly to deter unauthorized use of proprietary AI technologies. 10
- 11. #RSAC 11 Bias Scenario The algorithm of an AI hiring system employed by a tech company is discovered to disproportionately favor applicants from a specific demographic background. The bias stems from historical hiring data used to train the AI, which contains implicit biases against certain groups.
- 12. #RSAC Actionable Insights to Address Potential Bias Challenges Implement routine audits of AI algorithms to identify and correct biases. Develop a diverse training dataset that includes a wide range of demographics to reduce implicit biases. Establish clear guidelines and criteria for AI decision-making to ensure fairness. Create a feedback mechanism for applicants to challenge and review AI-driven decisions. 12
- 13. #RSAC Transparency Scenario A financial institution deploys an AI system for credit scoring. When applicants are denied credit, the institution cannot provide the reasons or mechanism for the decision due to the program’s complex decision-making process. 13
- 14. #RSAC Actionable Insights to Address Transparency Challenges Build in or enhance the AI system’s explainability, enabling it to provide clear reasons for credit decisions. Ensure compliance with consumer protection laws by documenting practices and disclosing criteria used for AI decision- making. Regularly review AI systems’ decisions for fairness and accuracy. Train customer-facing teams to explain decisions effectively. 14
- 15. #RSAC Reliability Scenario An AI system designed to predict machine failures in a manufacturing plant provides numerous inaccurate predictions, leading to unexpected downtimes and significant financial losses. The plant operators sue the AI system's providers for negligence, arguing that the providers failed to ensure the reliability of the AI system, which they rely upon for critical operational decisions. 15
- 16. #RSAC Actionable Insights to Address Reliability Challenges Implement rigorous testing and validation processes for AI system before launching. Establish contingency plans for operational failures, including manual overrides and regular maintenance checks. Ensure system providers carry liability insurance for potential failures. 16
- 17. #RSAC Liability Scenario An autonomous vehicle, while in full AI control mode, misinterprets traffic signals due to a software glitch and causes an accident. Parties involved: – The car manufacturer – The AI software developer – The vehicle owner 17
- 18. #RSAC Actionable Insights to Address Liability Challenges Clarify liability and insurance requirements in user agreements and terms of service. Develop standards for AI system performance, including safety and error-handling protocols. Implement a continuous monitoring and update mechanism for AI systems to prevent software glitches. 18
- 20. #RSAC The Legal Challenges of AI are Only in Their Infancy… Integrate ethical AI design principles. Consult with legal and compliance teams from start to finish in the program development or implementation. Implement rigorous testing and auditing. Adopt privacy by design. Assess and Prepare for liability exposure. 20