Innocence Project Statement in Response to the Biden Administration’s Initiatives to Address Risks and Opportunities Posed by Artificial Intelligence

The Innocence Project welcomes the Executive Order issued earlier this week by President Biden and the measures announced today by Vice President Harris to address the growing use of artificial intelligence (AI) in all areas of American life. We share the administration’s firm commitment to greater oversight and regulation of these new technologies. 

Over the past 31 years, the Innocence Project has freed or exonerated over 245 innocent people, identified the leading causes of wrongful conviction, and worked to transform the systems responsible for these injustices. Our cases have demonstrated that unreliable and inaccurate forensic technology is a leading cause of wrongful conviction. As a result, we are deeply concerned about law enforcement reliance on AI, especially when such technology is not thoroughly validated, tested, and regulated, before it is deployed. Indeed, we have already begun to see cutting-edge artificial intelligence-based technology – like facial recognition systems – drive wrongful arrests of innocent people.

We also recognize the power of AI and its potential to be a critical asset to all criminal legal system stakeholders. We therefore urge the Biden-Harris administration to establish a transparent and inclusive process for evaluating the reliability and validity of emerging and existing AI technologies, for assessing the ethical, legal, social, and racial equity implications of their use, and for developing technical and quality standards. During this process, the administration should seek input from the communities most directly affected by their proliferation and use. There must also be rigorous empirical analyses of AI tools before they are used by federal prosecutors and law enforcement agencies. These guardrails are necessary to prevent the wrongful conviction of innocent people.  

By proactively directing the development of standards governing the use of AI technologies, the Biden-Harris administration lays the groundwork for future research, development, and application that can and will guide the development of policies and practices that protect American security and privacy and advance equity and civil rights. We recommend the following:

• Preventing the use of facial recognition technology or predictive policing algorithms in the criminal legal system until there is sufficient research establishing their reliability and validity; allowing impacted communities the opportunity to weigh in on their scope and implementation; and mitigating algorithmic, racial, or societal bias.
• Pushing for more transparency around “black box” technologies whose inner workings are hidden from users.
• Mandating the disclosure of law enforcement use of AI technology in a criminal case to the defense so it can be subjected to adversarial testing in the courtroom.
• Making explicit the ways in which investigative technologies will be regulated to protect personal data. 

We welcome the administration’s leadership on this issue and acknowledge the importance of collaboration from all sectors — Congress, the private and philanthropic sectors, along with local and community organizations — in this endeavor. Only through these efforts can we protect innocent people from further risk of wrongful conviction in today’s digital age.