Advancing Government Services With Responsible Generative AI

The same goes for adoption of automated decision-making tools at the state and local levels. They’re used in law enforcement and the broader criminal legal cycle, in public benefit administration, in housing processes, and more. Certain states have pending legislation that would improve transparency and accountability of these tools state-wide, but none have passed yet.

Technology has created entirely new streams of data and platforms that law enforcement is being called on to police,47 posing the challenge of analyzing a virtually infinite amount of content with a very finite amount of human resources. Much like the case with content filtering, the law enforcement community views the new generation of AI-enabled tools as necessary to keep pace with their expanding technological purview. The NIJ recognizes this potential for AI, stating, “Examining the huge volume of possibly relevant images and videos in an accurate and timely manner is a time-consuming, painstaking task, with the potential for human error due to fatigue and other factors. Unlike humans, machine do not tire.”48 Beyond just its use in keeping pace with expanding amounts of content, AI can be used to provide more effective policing and crime prevention by detecting criminal warning signs earlier and apprehending suspects faster. Despite this risk, shared datasets are expected to become widespread within military AI operations. The DoD has already stated that the foundation for its AI efforts “includes shared data, reusable tools, frameworks, libraries, and standards…”39  The initial DoD AI applications, which focus on extracting information from aerial images and video, illustrate why sharing datasets is attractive.

Benefits of Conversational AI in the Public Sector

A second way to compromise data in order to execute a poisoning attack is to attack the dataset collection process, the process in which data is acquired. This effectively poisons the data from the start, rather than changing an otherwise valid dataset as shown in the example above. While physical attacks are easiest to think of in terms of objects, including stop signs, fire trucks, glasses, and even humans, they are also applicable to other physical phenomena, such as sound.

With the volume of personal information available and leveraged to train AI systems, we need to consider the potential risks and implications of these technologies. Google has a long history of driving responsible AI and cybersecurity development, and we have been mapping security best practices to new AI innovation for many years. Our Secure AI Framework is distilled from the body of experience and best practices we’ve developed and implemented, and reflects Google’s approach to building ML and gen AI powered apps, with responsive, sustainable, and scalable protections for security and privacy.

Artificial Intelligence (AL) and Machine Learning (ML) Security Best Practices for Local Government Systems

With Ask Sage, users can access a wide range of data sources in a natural language format, supporting teams in labor-intensive tasks, allowing them to focus on strategic initiatives. In addition, to strengthen cybersecurity, the EO directs that within 180 days, the Secretary of Defense and Homeland Security develop plans and conduct pilot projects deploying AI capabilities to identify and fix vulnerabilities in the US government systems. Reports on results and lessons learned are required to be submitted to the US President within 270 days. The EO focuses on developing guidelines, standards, and best practices for the safety and security of AI systems. Released on October 30th, 2023, the EO provides extensive directives that set the course for the federal government’s efforts to regulate the development and use of Artificial Intelligence in the US. With planning, government workforces can be augmented and empowered by conversational AI rather than displaced.

Research is being done to make this work by linking very specific data – potentially using various AI models and not just large language models – to do prompt engineering using specialized machine learning models on particular domains. In the face of these challenges, it is becoming increasingly clear that meaningful AI legislation is not guaranteed, and even Secure and Compliant AI for Governments if it does materialize, it may struggle to keep pace with technology’s rapid evolution. Therefore, a collaborative effort between the public and private sectors is crucial in safeguarding data privacy and promoting responsible AI development. One significant development on the horizon is the European Union’s (EU) AI Act, expected to be unveiled in 2024.