But, at least from a software perspective, keeping track of crypto is a lot harder than selling a Chagall. For that reason, the law enforcement agency has spent the past few years trying to hire a private tech company to help. But despite settling on contracts with crypto companies, at least two agreements appear to have fallen through. Today, the Marshals Service is still maintaining seized crypto on its own. 

“As the seizure and forfeiture of cryptocurrency has become commonplace, the USMS has sought to create a contract with private industry, just as it does with nearly all other asset types,” a spokesperson for the DOJ’s Asset Forfeiture Division told FedScoop. “Currently there is no private company that manages USMS’s cryptocurrency portfolio.” 

The search for a contractor started several years ago, when the US Marshals Service requested information from companies about the prospect of managing the agency’s cryptocurrency. In April 2021, a company called Bitgo, a crypto security company based in California, won a $4.5 million contract.

But, then, BitGo lost the agreement a few months after the Small Business Administration flagged the company as being too big to meet the contract eligibility. (Back in May, a company called Galaxy Digital had announced it planned to spend $1.2 billion to acquire BitGo, though the deal fell apart afterward.) In July, the Marshals Service hired another company, Anchorage Digital, which is based in San Francisco and also offers cryptocurrency holding services. 

Now, though, the Anchorage Digital contract also appears to have collapsed. As with the BitGo contract, the federal procurement data system shows that a Marshal Service contract with Anchor Labs was “terminate[d] for convenience.” Anchorage Digital is a subsidiary of Anchor Labs, according to its website. The company appears to have taken down a Medium post touting the agreement.

“Both awards were subsequently stayed pending the outcome of protests filed with the U.S. Small Business Administration (SBA), challenging the companies’ business size,” the USMS spokesperson told FedScoop. “Ultimately, SBA determined that both companies were other than small business.”

The company did not respond to a request for comment, though it’s worth noting that the Comptroller of the Currency issued a consent order against the company, which has an OCC banking charter, in 2022.  The Small Business Administration did not provide a comment by the time of publication.

“Not all cryptocurrency seized for forfeiture by the federal government is transferred to the USMS for custody and liquidation,” added the DOJ spokesperson. “The USMS utilizes the best practices and services of private industry to most effectively and securely manage and liquidate all assets in its custody.” 

The USMS has struggled with handling crypto, as a DOJ Office of Inspector General report highlighted last summer. At the time of the report’s publication, the Marshals Service was using multiple spreadsheets to manage its crypto, primarily because digital assets like bitcoin aren’t easily tracked in a DOJ property management program called the Consolidated Asset Tracking System (CATS).

These documents, according to the inspector general, don’t have “inventory management controls” and “documented operating procedures.” Policies for handling, storing, and valuing crypto are also “inadequate or absent, and in some instances provide conflicting guidance.” 

“The USMS’s supplemental spreadsheets do not have the capability to track edits made to the cryptocurrency entries in the USMS’s inventory records,” warned the inspector general. “As a result, these inventory records could be edited or deleted without a record of such a change being made and without the knowledge of individuals responsible for maintaining the spreadsheets.”

In some circumstances, the Marshals Service was “not fully complying” with rules for tracking crypto in CATS, the reported added.

The inspector general also said that the Marshals Service needs to develop more fleshed-out crypto policies before beginning work with a private company, cautioning that “without properly documented policies and procedures, the USMS lacks an adequate foundation for building performance requirements for a cryptocurrency services contract.”

The post Years later, the Marshals Service is still looking for help with seized crypto appeared first on FedScoop.

Speaking at a House Oversight Committee hearing last week, Ghost Robotics Chief Executive Gavin Kenneally said the ability of the technology to work in extreme environments while also detecting bombs, drugs and humans in need of help, could be a gamechanger.

He said: “The robots’ ability to traverse difficult terrain in all weathers and extreme temperatures makes it ideal for environments found along the U.S. southern and northern borders.”

Kenneally added: “Given that several hundred migrants die every year near the southern border from drowning or heat-related causes, a thermal camera could be equipped on the robot and used to identify them before it’s too late.

Philadelphia-headquartered Ghost Robotics supplies robots to the federal government and allied foreign governments. A few hundred of the company’s Vision 60 ‘robot dogs’ are already in use by the Defense Department for conducting perimeter security checks at Air Force bases and by law enforcement agencies.

Each robot costs $165,000 and can function in almost any terrain including rocks, sand, hills, ice, snow and staircases, and can operate in temperatures of -40 to 130 degrees Fahrenheit. According to Ghost Robotics, they can operate for over six miles on one charge at speeds of up to six miles per hour and store all data locally on computing devices that have undergone cybersecurity penetration testing. 

The device has multiple cameras and microphones built in and can interface with zoom thermal or infrared cameras to conduct more accurate analysis or anomaly detection of its surroundings before streaming video or information back to the human controller or a base station.

Some members of Congress including Rep. Nancy Mace, R-S.C., and Rep. Marjorie Taylor Greene, R-Ga., asked during the House Oversight hearing if there were ways to ensure that the robot dog or other innovative cutting-edge technologies can be stopped in the future from being used as weapons to hurt people physically or through cyber warfare.

Kenneally with Ghost Robotics said their robot dogs have no weaponry or harmful devices built into them and cannot take any action besides basic movement from one location to another without human input.

Furthermore, Wahid Nawabi, CEO of AeroVironment, a military drone and robotic systems company, said during the hearing that the devices his company makes are meant to use sensors with AI-driven algorithms to provide actionable intelligence for human beings like troops and engineers to make better decisions and are not meant to allow machines to make autonomous choices.

Ryan Rawding, the head of business development at Pangiam, a security and identity verification technology company for CBP and other agencies, added that his company creates software and tools using supervised machine learning models where the tools give risk-based scores to allow officers to make better decisions at the border while meeting CBP’s rules on data protections.

The post Ghost Robotics CEO: Robot dogs could save lives at US borders appeared first on FedScoop.

The hope is that the new group, which is expected to include leaders from the private and public sectors, will play a part in cultivating the technology — while also clamping down on its challenges.

The new working group will focus on several objectives related to generative AI, including developing new ways to evaluate and measure the technology’s effectiveness. The group will also help create guidance for using NIST’s AI risk management framework, which was crafted to inform the development of the technology.

Eventually, the group is expected to analyze how generative AI tools could help address some of the biggest challenges facing the country today, including climate change, according to a press release released by NIST on Thursday.

“This new group is especially timely considering the unprecedented speed, scale and potential impact of generative AI and its potential to revolutionize many industries and society more broadly,” Laurie E. Locascio, NIST director and Commerce undersecretary for standards and technology, said in a statement. “We want to identify and develop tools to better understand and manage those risks, and we hope to attract broad participation in this new group.”

The creation of the group represents he latest in the Biden administration’s AI agenda, which seeks to balance the opportunities and challenges created by the technology. The president met with AI experts earlier this week, and Senate Majority Leader Chuck Schumer announced his plan to regulate — and develop — AI on Wednesday.

The National Artificial Intelligence Advisory Committee, which was created by NIST last year, released its first report on Thursday, as well.

The post NIST launches public working group aimed at generative AI appeared first on FedScoop.

Brian Merrick, deputy director of solutions delivery staff for Justice, said during a webinar event the department’s Office of the Chief Information Officer is working through a draft policy involving privacy and the department’s application of AI technologies and “the considerations around using it.”

Privacy — as well as things like diversity, equity and inclusion — is “an active ongoing conversation that we make sure we circulate into any of the new emerging tech efforts so we’ve got the right controls in place, we’ve got the right equity holders involved and engaged fully, so that we make sure that we’re meeting all those requirements going forward, because obviously, being Department of Justice, we are highly focused on making sure that we follow those requirements,” Merrick said during the Federal News Network event.

While he couldn’t comment on when or if the draft policy might be made public, Merrick did say the AI privacy policy would be focused “in general [on] governing how we use the technology. But there will be certainly some intersections, I think, with the public interest, and when appropriate, we’ll make sure that the public interest is satisfied and all of our notification requirements.”

Interest in AI has exploded in recent months with the widespread introduction new capabilities like generative AI, and many federal agencies have begun exploring how they can take advantage of the emerging technology.

FedScoop recently spoke with Melinda Rogers, DOJ chief information officer, in an exclusive interview, during which she shared the department’s plans to use generative AI to improve customer experience for its IT service desk program.

In December 2020, the department issued an artificial intelligence strategy focused on
“cultivating an AI-ready workforce, aligning activities with the DOJ Data Strategy, building a governance structure, and supporting Department-wide AI adoption—with implementation designed to adapt to the evolving technology landscape.”

Merrick said the DOJ has “several AI efforts that are in play right now” including in the law enforcement community and for the legal community, particularly around “enhanced search options … [and] managing documents.”

“It’s a huge requirement for us as we’re one of the largest law firms, I guess, you would say,” he said.

Internally, Justice’s IT division is also using AI tools “with our own datasets that we manage to be able to glean those insights and help us expedite some of our processing,” Merrick said.

On the generative AI front, Merrick said “everyone is grappling with a completely different animal.”

“And so that’s gonna require a much more concerted effort as we really review policy with the rest of the world, frankly, and make sure that we’ve got our policy aligned with use cases and fully understand how that technology works,” he said. “So we’re still a bit off on that. But we’re starting to explore the possibilities and see what that looks like in the future.”

The post Justice Department developing privacy policy for AI appeared first on FedScoop.

The meeting was slated to include eight technology experts, including Jennifer Dounda, known for her pioneering work on gene editing, Khan Academy CEO and Founder Sal Khan, and former CEO of the Allen Institute for Artificial Intelligence Oren Etzioni.

“These experts include those who have been outspoken on the impact of AI on jobs, children, bias and prejudice, the risks posed by AI if it isn’t properly regulated, and also those who understand the benefits it provides for education and medicine if this technology is built safely from the start,” the White House official said in an emailed statement.

The Biden administration is exploring a regulatory approach to AI, as generative tools, like ChatGPT, have exploded in popularity in recent months. 

According to the White House official, the office of White House Chief of Staff Jeff Zients is supervising the development of “decisive actions” the administration can take over “coming weeks.”

The White House‘s work on the issue has also included discussions two to three times a week among White House principals and getting AI companies to commit to addressing public and private sector challenges, the official said. 

Other expected participants at the Tuesday event were:

• Tristan Harris, executive director and co-founder of the Center for Human Technology
• Jim Steyer, founder and CEO of Common Sense Media
• Rob Reich, professor of Political Science at Stanford University
• Joy Buolamwini, founder of the Algorithmic Justice League
• Fei-Fei Li, co-director of Stanford’s Human-Centered AI Institute

The post Biden meeting with technology leaders, academics on AI regulation appeared first on FedScoop.

In a request for information published Wednesday, State announced that a trio of its bureaus wants industry feedback about possibly using generative AI and machine learning capabilities to help with basic contract writing.

On top of this, State wants to glean “insight into the current hurdles and security considerations to introducing generative and natural language processing AI onto the Department’s network” through the RFI process.

Generative AI has exploded in popularity recently with the emergence of ChatGPT and other similar products that can generate new content, such as text or images, based on their training, which can include the use of large language models.

State’s bureaus of Information Resource Management — its CIO’s office — Consular Affairs, and International Narcotics and Law Enforcement are leading the effort.

“DOS business operations rely on outdated technology and manually intensive processes that result in unexploited data resources, wasted labor hours, and gross inefficiencies,” the RFI states. “The goal of embedding AI technology into an existing and recurring process is to reduce inefficiencies from manual laborious tasks, simplifying workflows, and improving the accuracy of repetitive tasks in the market research and acquisition planning phases while also addressing the nuances of IT-acquisitions.”

Currently, contracting officers typically copy and paste information from previous contracts to save time, State says in the document. But this can lead to errors or introduce risks, like “creating opportunities to exclude mandatory cybersecurity requirements while incorporating outdated provisions and clauses.”

The ideal solution would prompt a user to write a problem statement for the acquisition at hand and the generative AI solution would “generate a complete, draft PR package for any type of IT purchase, for a government procurement professional to review for potential edits, prior to submission in the contracting writing system,” the RFI explains.

The goal is this would not only reduce costs, manual labor and the chance of errors but also improve decision-making and deliver better contract outcomes, the department believes.

State admits there are some constraining factors that could limit moving forward with generative AI, including that the technology hasn’t been approved for use by the department, and the department doesn’t have a published AI policy yet. And as is the case with any federal RFI, the department is cl

Interested parties have until July 17 to respond to State’s RFI.

Secretary of State Antony Blinken is a major advocate for the U.S. being on the cutting edge of adopting emerging technologies like AI for global diplomacy. In January, Blinken kicked off operations of a new Office of the Special Envoy for Critical and Emerging Technology dedicated to the intersection of technology and diplomacy. And the department already has an extensive inventory of AI applications.

“We want the internet to remain a transformative force for learning, for connection, for economic growth, not a tool of repression,” Blinken said then. “We want to shape the standards that govern new technology, so they ensure quality, protect consumer health and safety, facilitate trade, respect people’s rights. We want to make sure the technology works for democracy, fighting back against disinformation, standing up for internet freedom, reducing the misuse of surveillance technology. And we want to promote cooperation, advancing this agenda tech by tech, issue by issue, with democratic partners by our side.”

The post State Department considers generative AI for contract writing appeared first on FedScoop.

But there’s a problem. The National Fire Incident Reporting System (NFIRS) — the government program used to track fires throughout the United States— is outdated. The tool, which originally relied on paper-based reports, was created by FEMA in the 1970s. The last time a new form of data was incorporated into the system was in 1995 when the Fire Administration started tracking Emergency Medical Services, or EMS, incidents. The system had its last technology update back in 2002. 

As a result, NFIRS is wonky and ill-equipped for studying and combatting fires of the 21st century. At this point, the system doesn’t allow fire departments to share information about the cause of the fires they encounter, which means the NFIRS only provides a limited amount of information. Another challenge is that only fire departments can access the NFIRS, blocking out other safety researchers from crucial data, said Michael Brooks, chief counsel for the Center for Auto Safety. 

“For me to get data, I have to pull it off [and] put it on a DVD to share it with you,” U.S. Fire Administrator Lori Moore-Merrell told FedScoop in a recent interview. “When Congress asks me something now, I have to say: ‘I can’t answer you for yesterday, but I can tell you 2021.’ So you see the problem.”

Now, the U.S. Fire Administration, which functions as a division of FEMA, is readying for a multi-year plan to completely overhaul its approach to fire incident reporting. The idea is to replace NFIRS with a new digital platform, announced in May, called the National Emergency Response Information System (NERIS). 

NERIS is supposed to be technologically sophisticated and will involve machine learning algorithms, a new app for firefighters, and a cloud-based interface that’s far more accessible. The goal is to make national fire incident data more robust — and more useful for dealing with the reality of climate change.

“A lot of the energy transitions happening across the U.S., whether it’s with electric vehicles or […] micro-mobility products, these high energy storage systems all carry with them a lot of unintended fire risk,” Moore-Merrell told FedScoop. “It’s also imperative that we can track wildfires’ movement.” 

But building NERIS won’t be easy. Only some U.S. fire departments submit data to the current NFIRS system, which means only about 70 percent of fires that happen in the U.S. are actually recorded, according to Birgitte Messerschmidt, the research director at the National Fire Protection Association. In some sense, NERIS may mean starting from scratch, since fire departments across the U.S. will need to be onboarded. Firefighters will require training on how to use the new app for sharing data with FEMA.

Notably, the platform will sit outside the federal government and be operated by an outside contractor called Fire Safety Research Institute, a nonprofit arm of UL Solutions, a safety science company based in Illinois. The technology is supported by a contract awarded through DHS’s Science and Technology Directorate program and has $2.2 million in funding for its first year. 

There’s also an ambitious timeline. The U.S. Fire Administration wants to have an initial system ready by early fall, start onboarding fire departments in 2024, and decommission the legacy program by 2025. Moore-Merrell told FedScoop that the program was already “a little behind anticipated,” while Messerschmidt, from the NFPA, called the schedule “optimistic.”

Still, both say the overhaul is critical. For example, the new data system could play a key role in tracking the fires created by batteries, a technology that will be key in stabilizing the electric grid and transitioning to electric vehicles. Right now, federal statistics on EV fires are relatively limited.

NERIS could also provide fire departments with new, real-time data to boost the efficacy of their responses more broadly. To make these improvements possible, NERIS will be set up to incorporate data sets from multiple federal agencies, such as data from the U.S. Census (including the American Housing Survey) and the National Weather Service. The system will also use Internet of Things components and physical sensors. 

But one of the most important parts of the system will be the new app for firefighters. This app, which is also being built by the UL Solutions nonprofit, will allow firefighters to directly input information about incidents from their phones and then send that information back to the NERIS database. The system may eventually incorporate biometrics, which will allow firefighters to volunteer to share health data — like information collected from smart watches — and support firefighting safety research. The Fire Administration also plans to use machine learning to scour emergency data for patterns, which could inform fire chiefs’ response to upcoming incidents. 

“If there’s a drought coming, or we’ve got weather forecasts saying, ‘Hey, it’s gonna be a particular red flag day,’ we can identify ‘Hey, there’s a higher chance of wildland fires so we might need to restructure,’” explains Craig Weinschenk, a principal investigator on NERIS at the Fire Safety Research Institute. He added that any algorithms the system uses will be open source. 

The government is slowly investing in ways to make better use of data collection, while also making that data more accessible. Eventually, researchers — and even Congress — are supposed to be able to request credentials so that they can access the NERIS system online, too. 

In the meantime, Moore-Merrell said that the Fire Administration has not yet settled on a cloud provider for the system.

Time will tell whether FEMA will succeed in building NERIS — and according to schedule. Still, the new software is evidence that the government is investing in ways to make better use of data collection, while also making that data more accessible. 

“What’s becoming far more common is for there to be online data aggregators that pull together several of the raw data sources together and also serve them up visually, in usually a map format to the public,” explains Kendra McLauchlan, a National Science Foundation program manager who focuses on wildfires. “Often, the case is that the raw data are in forms that cannot really be easily seen or used by the public.”

The post FEMA plans new fire tracking system armed with machine learning and biometrics appeared first on FedScoop.

The new service will allow government departments to adapt models including GPT-3 and GPT-4 for specific tasks, including content generation, summarization, semantic search, and natural language-to-code translation.

The language models will run within Microsoft’s cloud service for U.S. government agencies, Azure Government.

“If you’re an Azure Government customer (United States federal, state, and local government or their partners), you now have the opportunity to use the Microsoft Azure OpenAI Service through purpose-built, AI-optimized infrastructure providing access to OpenAI’s advanced generative models,” Bill Chappell, Chief Technology Officer, Strategic Missions and Technologies at Microsoft said in a blog post shared with FedScoop.

“Microsoft has developed a new architecture that enables government agencies to securely access the large language models in the commercial environment from Azure Government allowing those users to maintain the stringent security requirements necessary for government cloud operations,” Chappell added.

Notably, Microsoft says all traffic used within the service will stay entirely within its global network backbone and will never enter the public internet. The technology giant’s network is one of the largest in the world and made up of more than 250,000 km of lit fiber optic and undersea cable systems.

The tech company added that the Azure OpenAI Service does not connect with Microsoft’s corporate network, and that government agency data is never used to train the OpenAI model.

The Azure OpenAI Service can be accessed using REST APIs, Python SDK, or Microsoft’s web-based interface in the Azure AI Studio, and all Azure Government customers and partners will be able to access all models.

Microsoft is doubling down and highlighting its data, privacy, and security protections offered to government customers by encrypting all Azure traffic within a region or between regions using MACsec, which relies on AES-128 block cipher for encryption. 

The post Microsoft launches generative AI service for government agencies appeared first on FedScoop.

The proposed program is called the Track Three Pilot Program and would allow “micro entity” applicants — that status is governed by federal rules — to take a 30-month period before paying search or examination fees, provided they meet certain requirements. Applicants could also receive a “obtain a pre-examination search report” before paying an examination fee.

“The USPTO recognizes that under-resourced applicants may need a low-cost option with minimal requirements to allow them additional time for commercialization efforts and to ascertain the value of their inventions,” explained Katherine Vidal, the undersecretary of commerce for intellectual property and the director USPTO  in a post shared to the Federal Register on Friday. 

The pre-examination search report option that the office is considering could involve artificial intelligence. Recently, USPTO incorporated an AI-enhanced tool —which was trained on past patent data —  that allows examiners to analyze how similar an application is to previously-filed domestic and foreign patent documents. Now, the office is considering whether the search results produced by that AI system should be included in a pre-examination report.

The goal, the federal register post said, is “to provide applicants with additional information as they consider potential commercialization and the value of their invention.”

The move comes as the office looks to use AI to accelerate the patent application process. The comment period closes on July 25. 

The post AI could have new role in speeding up US patent process appeared first on FedScoop.

“I sat down with a bunch of contracting officers and said I’m going to build a bot. And I had people look me dead in the eye and say, ‘I will never trust a bot,'” Kinnard, Labor’s director of innovation and engineering, said at the UiPath Together Public Sector summit.

Fast forward to today, and now 140 people use that bot on a daily basis and save “hundreds of thousands of dollars every year” in doing so, she said. And under Kinnard’s leadership, the department has further expanded its RPA work to other areas like HR processing, hiring, and other administrative tasks.

How was Kinnard able to make believers and willing users out of skeptics? “You have to bring people along on the journey,” said Kinnard, who has led emerging technology at DOL for a number of years and was recognized by the Partnership for Public Service’s Service to America Medals program for her work with RPA.

“These contracting officers go to be part of our development process,” from the early steps to build the bot to working through when it didn’t work the first time, she said. “And now they are the biggest champions of automation. They are not only coming to me saying let’s build more bots, they are going to the rest of the organization and saying, ‘Hey, we’ve got this really great bot, you guys need to get a really great bot too.'”

Kinnard said government agencies like Labor are “at this really great nexus where the culture is finally ready to say yes, bots are an amazing tool to help us get our job done.”

And there’s an opportunity to build off of the increased appetite for bots, — particularly as federal agencies innovate and pilot robotic process automation use cases — to share and connect applications across agency borders, she said.

“There is the really great opportunity to say: ‘You have a system over here, I have a system over here. Let’s really make the systems the most powerful they can be by letting them talk to each other through automation,'” Kinnard said. “That’s really where I see the next few years of automation going to next.”

The post ‘I will never trust a bot’: How the Labor Dept. overcame cultural resistance to RPA appeared first on FedScoop.