Congress Explores AI Use and Confronts Security Risks

Introduction

With over 100 congressional offices already utilizing artificial intelligence (AI) for various tasks, Congress is eager to explore more opportunities to leverage this technology. The aim is to alleviate the workload of overwhelmed staff members, assist with research, draft legislation, and enhance constituent outreach capabilities. Congress recognizes the need to build staff capacity without expanding payroll and is determined to keep up with technological advancements.

Building Guardrails for AI Use

Congress is actively working on establishing early guardrails for AI implementation. The House's Chief Administrative Office (CAO) plans to introduce a draft policy for AI use across the House within the next few months. Deputy CAO John Clocker emphasized the transformative potential of AI but also stressed the importance of caution and appropriate safeguards to protect against potential risks. Adversaries could exploit AI tools to harm the House, making it crucial to establish guidelines and frameworks.

The House's Approach

The House's AI policies will be based on the National Institute of Standards and Technology's AI Risk Management Framework. However, the CAO will tailor these policies to accommodate the unique ecosystem within the House. While the House may adopt broad guardrails for AI usage, individual members and their offices will have the flexibility to manage AI implementation based on their appetite for innovation and risk.

Lawmakers' Vision for AI

Lawmakers already have ideas for how AI can benefit them and their staff. For example, Rep. Morgan Griffiths envisions being able to listen to audio versions of reports or bill text during his commute to Washington. Rep. Norma Torres wants AI to assist her district staff in managing overwhelming amounts of constituent casework while ensuring the protection of personal information.

Early Adoption of AI in Congress

While the use of AI in Congress is still in its early stages, it mirrors the broader trend of major AI programs being developed outside of the public sector. Congress's plans to leverage AI represent a rare example of adopting a technology that is still being refined and developed. However, it is important to acknowledge that AI is not without its flaws, with AI systems exhibiting inconsistencies and overconfidence.

Current Use of AI in Congress

Currently, more than 200 staffers in 150 House offices, including committees, are participating in a pilot program using Chat GPT+ for everyday tasks such as scheduling, constituent correspondence, and bill summaries. The most common use of ChatGPT+ is to generate initial drafts of testimony, statements, or speeches, which are then edited by staff members to incorporate the necessary voice and style that AI cannot yet achieve.

Choosing an AI Provider

While the House has explored other generative AI platforms like Google's Bard and Microsoft's Copilot, OpenAI's ChatGPT is the only company that has committed to protecting House and member data. OpenAI has pledged not to use the data in training its models or share it with other customers. The CAO's office continues to evaluate other providers, but currently, the paid license version of ChatGPT+ remains the only AI provider approved by the House.

The Senate's Progress with AI

The Senate is not as far along in its experimentation with AI use. However, a working group was established last year, and some guidance has been provided to offices involved in pilot efforts. The Senate's cybersecurity officials have determined that OpenAI's ChatGPT, Google's BARD, and Microsoft's Copilot present a moderate level of risk if proper controls are followed. Currently, the Senate restricts AI use to research and evaluation purposes, with non-sensitive data.

Conclusion

Congress is actively embracing AI and exploring its potential applications. While security risks must be addressed, lawmakers are determined to strike a balance and leverage the benefits of AI while protecting against any downsides. With the House leading the way in establishing guardrails and the Senate making progress with pilot programs, Congress is positioning itself to navigate the ever-evolving digital landscape.

—————————————————————————————————————————————
By: Katherine Tully-McManus
Title: Congress Explores AI Use and Confronts Security Risks
Sourced From: www.politico.com/news/2024/01/30/congress-security-artificial-intelligence-00138643
Published Date: Tue, 30 Jan 2024 16:53:08 EST