The Wrap: DISA Data Strategy Incoming; GenAI Evaluations Ongoing; AISI Eyes GenAI Frontiers

Welcome to The Wrap for Wednesday, June 26!

?

From the newsroom at MeriTalk, it’s the quickest read in Federal tech news. Here’s what you need to know today:

?

DISA Data Strategy Incoming

The Defense Information Systems Agency (DISA) is putting the finishing touches on a new data strategy for 2025-2027 that the combat support agency hopes to release to the public in the coming months. That was the news from DISA Chief Data Officer Caroline Kuharske who briefed reporters on Tuesday at AFCEA International’s TechNet Cyber conference in Baltimore. “With the previous [data strategy], we really wanted to get our hands or get our head around what data do we even have, what pockets do we have, what silos do we have,” Kuharske said. “Now that we know what we have, we have the data catalog. Now we’re going to start doing more integration of data governance and policies in the backend of systems.” Putting the new strategy in place will ensure that whenever a “program application comes through different governance boards within the agency that they that they are applying the data governance and policies” put out by DISA’s CDO office, Kuharske said. “[We’re] going from just trying to figure out what the lay of the land is to operationalizing it through the different technologies that we can use,” she said.

?

GenAI Evaluations Ongoing

While many Federal agencies are at least testing the generative AI waters, one of the biggest of those – the Department of Defense (DoD) – is putting a premium on continual testing and evaluation of large language models to measure their value and safety. Speaking today at the AWS Summit?in Washington, Pentagon Chief Digital and Artificial Intelligence Officer (CDAO) Radha Plumb said her office is continually testing and evaluating “to make sure the models are doing what we want,” and to make sure GenAI models are aligned with the DoD’s?Responsible Artificial Intelligence (RAI) Toolkit. “I think having test and evaluation, having these ongoing conversations where on the government side we can understand where the models can and should be tested – both by the companies and then inside the government – and building that in as part of adoption is part of what proving the value proposition of responsible AI globally is,” Plumb said.

?

AISI Eyes on the Frontier

Also speaking at the AWS Summit today, Elizabeth Kelly, director of the National Institute of Standards and Technology’s (NIST) recently established?U.S. AI Safety Institute?(AISI), said her organization is focusing on AI safety but especially regarding "frontier" GenAI models. What are those? According to the AI app just employed by The Wrap to answer that question, frontier AI models are ones on the “cutting edge of artificial intelligence, so powerful they're considered to be pushing the boundaries” of what AI can do. “We’re going to be building out a suite of evaluations to use data on how AI models perform, what capabilities they exhibit, and what risks are posed by those capabilities,” Kelly said today. “This is going to be an entirely new U.S. government capacity to directly test frontier AI models and systems before deployment across a range of risks.” She continued, “in our initial pilot stage, we’re focused especially on capabilities that could pose a threat to national security.” And she said AISI will look at whether these models could be used to perpetuate more effective cyberattacks or to develop biological weapons. “We’re also going to be looking at societal harm perpetuated by frontier models and systems,” Kelly said. “And we’re going to be sharing feedback with model developers on where mitigations may be needed prior to deployments.”

?

Cyber Workforce Legislation Tease

Rep. Andrew Garbarino, R-N.Y., chairman of the House Cybersecurity and Infrastructure Protection Subcommittee, dropped heavy hints at a House Homeland Security Committee hearing today that he and committee Chairman Mark Green, R-Tenn., are cooking up legislation to address the U.S. cybersecurity workforce shortage, and plan to offer the legislation “soon.” No firm news to report on the contours of a possible bill, but today’s hearing certainly gave shape to the problem. “Experts predict that by the end of 2024, a cyberattack will strike every 13 seconds. That’s 6,822 attacks a day, or about 2 million by the end of the year,” said Rep. Green. “It is alarming, then, that our nation is suffering from such a massive cyber workforce gap. We currently need at least 500,000 cyber professionals if we hope to protect and defend our way of life.”

?

Once again, let’s “call IT a day,” but we'll bring you more tomorrow. Until then please check the MeriTalk breaking news website throughout the day for the latest on government IT people, process, and policy. And finally, please hit the news tip jar [with leads, breaking news, or simply your two cents] at [email protected].