The Department of Homeland Security has seen the opportunities and risks of artificial intelligence firsthand. It found a trafficking victim years later using an A.I. tool that conjured an image of the child a decade older. But it has also been tricked into investigations by deep fake images created by A.I.
Now, the department is becoming the first federal agency to embrace the technology with a plan to incorporate generative A.I. models across a wide range of divisions. In partnerships with OpenAI, Anthropic and Meta, it will launch pilot programs using chatbots and other tools to help combat drug and human trafficking crimes, train immigration officials and prepare emergency management across the nation.
The rush to roll out the still unproven technology is part of a larger scramble to keep up with the changes brought about by generative A.I., which can create hyper realistic images and videos and imitate human speech.
“One cannot ignore it,” Alejandro Mayorkas, secretary of the Department of Homeland Security, said in an interview. “And if one isn’t forward-leaning in recognizing and being prepared to address its potential for good and its potential for harm, it will be too late and that’s why we’re moving quickly.”
The plan to incorporate generative A.I. throughout the agency is the latest demonstration of how new technology like OpenAI’s ChatGPT is forcing even the most staid industries to re-evaluate the way they conduct their work. Still, government agencies like the D.H.S. are likely to face some of the toughest scrutiny over the way they use the technology, which has set off rancorous debate because it has proved at times to be unreliable and discriminatory.
Those within the federal government have rushed to form plans following President Biden’s executive order issued late last year that mandates the creation of safety standards for A.I. and its adoption across the federal government.
The D.H.S., which employs 260,000 people, was created after the Sept. 11 terror attacks and is charged with protecting Americans within the country’s borders, including policing of human and drug trafficking, the protection of critical infrastructure, disaster response and border patrol.
As part of its plan, the agency plans to hire 50 A.I. experts to work on solutions to keep the nation’s critical infrastructure safe from A.I.-generated attacks and to combat the use of the technology to generate child sexual abuse material and create biological weapons.
In the pilot programs, on which it will spend $5 million, the agency will use A.I. models like ChatGPT to help investigations of child abuse materials, human and drug trafficking. It will also work with companies to comb through its troves of text-based data to find patterns to help investigators. For example, a detective who is looking for a suspect driving a blue pickup truck will be able to search for the first time across homeland security investigations for the same type of vehicle.
D.H.S. will use chatbots to train immigration officials who have worked with other employees and contractors posing as refugees and asylum seekers. The A.I. tools will enable officials to get more training with mock interviews. The chatbots will also comb information about communities across the country to help them create disaster relief plans.
The agency will report results of its pilot programs by the end of the year, said Eric Hysen, the department’s chief information officer and head of A.I.
The agency picked OpenAI, Anthropic and Meta to experiment with a variety of tools and will use cloud providers Microsoft, Google and Amazon in its pilot programs. “We cannot do this alone,” he said. “We need to work with the private sector on helping define what is responsible use of a generative A.I..”
