That’s the premise behind the Computing Community Consortium’s (CCC’s) session during the CRA Summit at Mystic Lake Center. On Thursday, July 23rd, members of the CCC Council Executive Committee will host Yes, And: Computing Visioning Improv, a session that trades a polished panel presentation for spontaneous, crowdsourced ideas. The session aims to bring researchers from across various fields together to do some off-the-cuff visioning, imagining in unusual ways what the field could be capable of long-term.

A New Computing Card Game

The format is built around a card game CCC has developed specifically for this kind of creative brainstorming. In small teams, participants will draw from two decks: one featuring computing substrates like quantum computing or biological computing, and one featuring application domains ripe for transformation, from massive undertakings like agriculture to something as mundane as email. The challenge is to combine whatever cards you draw and pitch a compelling vision for how that substrate could address a real problem in the application.

The crowd serves as both audience and judge, evaluating whether each idea is bold enough, worthy of investment, and pointed at the challenges that matter most. And everyone attending is encouraged to participate, not just watch.

Join the CCC Session

This session is about more than just generating interesting — or wacky — ideas. CCC aims to encourage the kind of visionary thinking that doesn’t always fit neatly into a grant proposal, and to create space for researchers across disciplines to actually talk to each other. The card game format is deliberately designed to produce unlikely combinations because those collisions tend to be where we’re most challenged to think out-of-the-box. In this way, the visioning happening during Yes And: Computing Visioning Improv is a direct reflection of CCC’s mission to shape the future of computing research. All CRA Summit attendees are welcome, whether as a participant or an observer.

Read more about the CRA Summit here.

Tune in to the CCC LinkedIn Showcase Page for updates like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

AAMAS is the largest conference in the area of agents and multiagent systems, bringing together researchers and practitioners in all areas of agent technology and providing a forum for publishing and finding out about the latest developments in the field. AAMAS is the flagship conference of the non-profit International Foundation for Autonomous Agents and Multiagent Systems (IFAAMAS).

AAMAS 2026 Blue Sky Winners

First Place

Foundation World Models for Agents that Learn, Verify, and Adapt Reliably Beyond Static Environments

Florent Delgrange 

Equal Runners-Up

Guiding Sociotechnical Systems toward Value-Norm Equilibrium

Nirav Ajmeri, Marina De Vos, Davide Dell’Anna, Pradeep K. Murukannaiah, Vivek Nallur, Luis Gustavo Nardin, Munindar P. Singh

The Dynamic Turn in Strategy Logics

Rustam Galimullin, Maksim Gladyshev, Munyque Mittelmann, Nima Motamed

Please join us in congratulating the AAMAS 2026 winners! 

CCC sponsors many Blue Sky tracks at various conferences in computing every year. If you are interested in requesting sponsorship of a Blue Sky track at your conference, please email blueskyideas@cra.org. 

Requests need only include a brief description of the conference and a proposed list of program committee members for the track. For more information — including guidelines for conference program committees, recommendations for selecting winners, and logistics for issuing CCC-sponsored travel awards to the winners, as well as a sample call for papers for a Blue Sky Ideas track — visit our website.

Tune in to the CCC LinkedIn Showcase Page for updates and more Blue Sky track announcements. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

Visioning is at the core of what the Computing Community Consortium (CCC) does, but it can be a nebulous concept. What does it mean to actually “do visioning”? Why does it matter for computing research? And where does CCC come into the picture?

Our new Visioning in Action series aims to answer those questions and more by providing a glimpse into the behind-the-scenes of our current visioning initiatives. Learn about how our workshops, reports, Community Chats, and more come to be through the words of community members who work with us. 

For the first part of this series and to learn more about the Defining the Role of Computing Interfaces in Neural Interface Design visioning initiative, check out this CCC blog post. There, you can also read about what the proposal process for organizing a CCC workshop looks like.

Organizing a Multi-Field Workshop

The Defining the Role of Computing Interfaces in Neural Interface Design workshop took place April 22-23, 2025, about one year after the acceptance of the final proposal for the workshop. For co-organizers Abhishek Bhattacharjee (Yale University), Raghavendra (Raghav) Pothukuchi (The University of North Carolina at Chapel Hill), and Nishal Shah (Rice University), much of that time was spent on tasks like deciding who to invite, developing small group discussion questions, and selecting speakers to energize the conversation at the workshop. In addition to managing logistics like location, registration and invitations, meals, and equipment, CCC staff also shared the institutional knowledge that CCC has gathered over the years to help organizers develop a well-rounded, engaged, and fruitful workshop.

According to Abhishek, the most challenging aspect of organizing the workshop was creating a list of invitees. To produce a meaningful understanding of the future of computing in brain-computer interfaces (BCIs), the workshop needed a wide variety of experts, including neuroscientists, bioethicists, regulatory agents, and, of course, computing researchers across domains. And more than just representing their field, these individuals had to be a fit for visioning itself — to be willing to work collaboratively with such a diverse group and to think boldly about the future of the field.

While considering the various perspectives that may enrich the discussion and developing the invite list, the organizers were even fortunate enough to connect with a few individuals with implanted BCI devices. “Many bioethicists get their cues from the actual individuals, and we had a few people at the workshop who themselves had these implants… it’s one thing to have neurosurgeons sort of present to computing people on behalf of those patients, but it’s another thing entirely to get those individuals in the same room as the computing folks,” said Abhishek.

The Workshop Comes Together

During the actual workshop, the organizing committee helped lead the proceedings with support from CCC staff. The agenda blended thought-provoking speakers and Q&As with breakout sessions designed to prompt open-ended discussion in small groups over two days. 

At first, there was some settling in as folks from different domains got comfortable with each other’s research terminology. But ultimately, the conversations flowed freely and illuminated the numerous considerations going into computing in BCIs. The organizing committee noted that, in a way, the workshop went against what is typically viewed as a “success” in technology. Success is often defined as not necessarily needing to have a full understanding of how all other aspects of a technology work for a specialist — in hardware, software, or neurosurgery, for example — to expertly do their specific job. Rather, at this workshop, participants were able to share their perspectives and listen, learning about each other’s work with a goal of advancing BCI technology. 

By convening folks from across fields, with different accepted jargon and research areas, the workshop ran the risk of becoming too broad and losing sight of what the actual goal was: defining the role of computing, specifically, in shaping the future of BCIs. From Nishal’s perspective, “One of the things that Abhishek and Raghav did well was to remind people that it’s about computing at the end. Because there were so many diverse topics, we could just get lost, and everyone had their own specialties. So, it was important to remind them that there was a central goal: computing.”

Finally, Abhishek highlighted the particular value of having CCC staff live onsite to assist with running the workshop and adapting to last-minute changes. “We didn’t feel comfortable cutting speakers off, because the Q&A… signifies engagement, and we blew over the allotted budget significantly,” he said. “What helped was [CCC staff] telling us this was normal, and [giving us] a plan to redo the other sessions so that by the end of the day, we were still on schedule.”

Shaping Future Research

After bringing the computing community (and beyond) together at the workshop, Abhishek, Raghav, and Nishal pivoted to developing a report with findings and recommendations informed by the conversations that took place. For Raghav, “At the end of the day, when this report gets sent out, it has the implication that it’s a field consensus. And we [made] sure that the discussions that we had at the workshop carried that weight.”

Being so comprehensive, however, means that the organizing committee needs to keep in mind the same challenge that the workshop discussions themselves had: staying within scope. Doing justice to the multitude of topics that arose has been the biggest challenge during the report-writing process, but is a testament to their commitment to making a real impact through this visioning initiative.

Through this report, the organizing committee hopes to guide future research directions as well as make the case for funding in areas that will support computing applications in BCIs. But more than that, they hope to create a future for this nascent domain. 

“Computing for BCIs doesn’t really exist wholesale… I hope [the report] empowers senior people to feel that they can make more grounded claims in the importance of this work, so that junior practitioners are given the long runway which they need to succeed and for the field to succeed,” Abhishek said. 

Stay Tuned for the Full Workshop Report and Get Involved with CCC Visioning

If you’re interested in learning more about how to get involved with CCC visioning initiatives or want to propose one yourself, we encourage you to read more about visioning proposals here. For the full scope of what it’s like to organize a visioning initiative with CCC, you can read our Visioning Best Practices document.

The full workshop report for Defining the Role of Computing Interfaces in Neural Interface Design will be released in Summer 2026, laying out key findings and strategic recommendations for the future of computing in neural interfaces. Subscribe below or follow us on LinkedIn to be the first to know when it’s released.

Tune in to the CCC LinkedIn Showcase Page for updates and more Visioning in Action. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

Watch the Recording Here

Both the Community Chat and the CCC workshop report — co-authored by Lucy Fortson (University of Minnesota), Lea Shanley (International Computer Science Institute, Berkeley), Tanya Berger-Wolf (The Ohio State University), Kevin Crowston (Syracuse University), Corey Jackson (University of Wisconsin-Madison), and Saiph Savage (Northeastern University) — center around two fundamental questions: why is the convergence of citizen science and advanced computing needed, and how do we make it happen on a large scale?

The Need for Convergence

There are a number of gaps in the current scientific research ecosystem that converging computational and citizen science research can help fill. Corey Jackson presented the findings of the report which aims to answer the two above questions. 

1. Mutual reinforcement between AI/ML and citizen science

There is an underutilized feedback loop between AI/ML and citizen science, where each can train and improve the other. 

In short, as Jackson explained, “Citizen science gives contextualized training data, and AI gives citizen science real-time coaching and the ability to scale up.” 

2. Human-computer teaming is essential

Bringing citizen science and computational science together makes the most of the unique strengths of humans and computers while helping mitigate their weaknesses.

“Even the best models drift. They generalize poorly across instruments and oftentimes miss context… Humans can catch what the AI misses and the AI can handle what humans can’t do at scale.”

3. Feedback and interactivity are critical for engagement

“Feedback is a gap that is largely treated as an afterthought,” Jackson said. Despite this, personalized, real-time, and reciprocal feedback drives participant retention and helps ensure data quality.

4. Trust is fragile but foundational

Jackson made sure to highlight the importance of user trust in citizen science research projects, and how all other factors need to be addressed to help maintain that trust.

“Opaque algorithms, inaccessible interfaces, they all kind of erode the willingness of volunteers to contribute and the willingness of scientists to use the data.”

5. Infrastructure as bottleneck and opportunity

Out-of-date infrastructure hinders the success of citizen science, but revitalizing it promises a new capacity to organize increasingly global research projects that handle larger amounts of and more complex data.

“Most citizen science platforms launched around 2006, the same era as the smartphone…” Jackson noted. “Today they’re running globally on devices ranging from drones to low-cost sensors, and in places with intermittent connectivity. The infrastructure hasn’t caught up.” 

6. Rising security, privacy, and adversarial threats

Data poisoning, synthetic media, and coordinated disinformation all pose threats to citizen science research. 

“As citizen science feeds into agency decision-making — NASA, NOAA, for example — it becomes a target… The defenses we [have] are underdeveloped.”

7. Momentum is real but fragmented

Finally, there can be a certain instability around citizen science research projects that hinders their potential. Currently, there is a lack of shared standards, sustained funding, and coordinated governance that would help these projects succeed long-term.

“We have a ton of wins. There are a ton of really great projects. But we really need to be thinking about how we share knowledge to make these projects and these approaches more sustainable,” Jackson urged.

A Research Roadmap for the Next Decade

To address those gaps, the co-organizers synthesized findings from the workshop into five key research drivers:

• Human-Machine Teaming: How humans and machines should divide labor and research workflow.
• Feedback, Interactivity, and JIT Delivery: The full feedback loop between volunteers, project teams, and society.
• Trust, AI, and Citizen Science: Building a shared future on transparency, governance, and measurable trust.
• Security, Privacy, and Open Systems: Keeping participatory systems both open and secure.
• Future Infrastructure: Foundational, sustainable infrastructure across cyber, data, human, and physical layers.

These research drivers shaped the recommendations presented in the report for the future of convergence. 

“[The question was] not what we can do in the next year, but what the field can look like in ten years if we invest deliberately now.” — Corey Jackson

Citizen Science Across Domains

The discussion was also enriched by the addition of two unique lightning talks showcasing the value of citizen science in different scientific domains — and how it intersects with advanced computing to create that value. 

The first lightning talk was presented by report co-author Tanya Berger-Wolf, who presented about how ecological citizen science research has already helped push the boundaries of computing research. “The bulk of nature data today comes from citizen science contributions,” Berger-Wolf explained, and those contributions provide large, ML-ready datasets that provide the foundations for some computing research. “These datasets are the biggest source of AI and ML benchmarks, especially image-based ones.”

Berger-Wolf was followed by Marc Kuchner, Citizen Science Officer at NASA. Kuchner highlighted another research dynamic already taking place at NASA: citizen scientists writing and using their own AI tools. One NASA volunteer, for example, developed a new AI tool, uniquely suited to the research project he was involved in, to help identify clouds produced by pollution from rockets. At the same time, he pointed out that researcher trust could be just as fragile as volunteer trust in citizen science projects: “With AI/ML, anyone can be a scientist, or pretend to be one. That makes educating and guiding the public via real citizen science even more important, rewarding, and challenging.” 

Watch the Webinar and Read (and Cite!) the Report

If you missed the event or want to revisit the discussion, you can watch the full recording above and read the report here:

Grand Challenges for the Convergence of Citizen Science and Computational Research 

We encourage you to share the report and webinar recording widely with your colleagues, collaborators, and especially with students and early-career researchers who are shaping the future of citizen science and computing research.

Tune in to the CCC LinkedIn Showcase Page for updates and more Community Chats like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

In this new brief, Key Takeaways for the Convergence of Computational and Citizen Science Research, read the essential findings and recommendations from the full workshop report in an abbreviated form. The brief covers three primary areas of recommendations:

• National Infrastructure for Convergence: The sustained platforms, governance systems, and physical/cyber architecture required to support scalable, trustworthy, and nationwide convergence efforts.
• Core Research for Convergence: The foundational scientific and socio-technical investigations required to advance the field, focusing on developing new models, metrics, and frameworks for human-AI interaction, trust, and accountability.
• Training and Capacity Building: Developing the human capital — the skills, knowledge, and organizational structures — needed to create, manage, and participate in convergence projects across all sectors.

Read the Brief Now

View the Key Takeaways for the Convergence of Computational and Citizen Science Research brief below. You can also read the full Grand Challenges for the Convergence of Computational and Citizen Science report here.

Read the Brief Here

Learn More at the Upcoming Community Chat

For another way to engage with the findings in this report, join us for an upcoming CCC Community Chat on May 20th, 2026 at 3:30pm. This 45-minute Zoom webinar, hosted by the report’s authors, will dive into key findings, follow up with a series of lightning talks with experts from citizen and computational science, and end with a Q&A. Featuring guest speaker Marc Kuchner, Citizen Science Officer at NASA!

Register and Submit Questions Here

Visioning is at the core of what the Computing Community Consortium (CCC) does, but it can be a nebulous concept. What does it mean to actually “do visioning”? Why does it matter for computing research? And where does CCC come into the picture?

Our new Visioning in Action series aims to answer those questions and more by providing a glimpse into the behind-the-scenes of our current visioning initiatives. Learn about how our workshops, reports, Community Chats, and more come to be through the words of community members who work with us.

Getting to Know Defining the Role of Computing Research in Neural Interface Design

In April 2025, CCC hosted the Defining the Role of Computing Research in Neural Interface Design workshop in Washington, D.C.. The workshop was co-organized by Abhishek Bhattacharjee (Yale University), Raghavendra (Raghav) Pothukuchi (The University of North Carolina at Chapel Hill), and Nishal Shah (Rice University), with support from CCC Council Member Weisong Shi (University of Delaware), Jojo Platt (Platt and Associates), and CCC staff. 

In order to better understand the role of computing in neural interfacing, the workshop convened researchers across computing and neuroscience, representatives from regulatory agencies, industry representatives, and individuals living with contemporary neural interfaces to assess the long-term challenges facing these complex pieces of technology. Through their discussions, participants aimed to create a strategic roadmap for tackling those challenges and pushing innovation in neural interfacing forward — in other words, to “vision” for the domain. 

Co-organizers Abhishek, Raghav, and Nishal graciously agreed to tell us more about what the visioning initiative organizing process looked and felt like on their end.

The Initial Spark

The first step of organizing a CCC visioning initiative is to submit an interest form. This expression of interest is thorough — asking about the different perspectives that could inform the topic, what activities are being proposed, and desired outcomes, for instance — but can convey the very beginning stages of a visioning idea. While it should be well-considered, there is no expectation that it reflects the final scope of the topic. For Abhishek, Raghav, and Nishal, their initial motivation for putting together a visioning workshop was to bring together researchers across the sciences to improve brain-computer interfacing (BCI). 

“It had become clear through the process of writing papers [on this topic] that if you want to build a computer system for BCI, you can’t just focus on one silo of the computer design,” Abhishek said of their initial thought process “…You need to think about the circuit-level issues, wireless and communication issues, the system software. There are many pieces that we don’t have answers to.”

This has been a particular challenge historically because computer scientists, Abhishek and Raghav noted, have been isolated from the biological side of this research. Raghav explained that, “This is a theme which requires very tight conversations between neuroscientists and computer scientists — people who do not usually meet with each other. Even if you look at a lot of university campuses, a lot of times, the medical school is far off, and engineering and computer science are elsewhere.” To that end, a visioning workshop provided an excellent opportunity to bring everyone to one table and assess what’s needed for the future of BCIs as well as how to address those needs.

Collaboration with CCC

All three organizers agreed, however, that the scope of the workshop shifted and expanded during the next step in the process: feedback from the CCC Council. Abhishek said that, ultimately, that feedback meant the workshop they organized ended up being pretty different than the one they had originally imagined. Throughout the proposal process, the Council encouraged them to think about computing needs more broadly, to really think about the ethical implications of BCI work and include those voices in the conversation, and make sure to keep in mind the societal implications of this research.

Nishal noted that the audience scope, in particular, was widened after discussion with the Council. “[We were thinking of] releasing data on a data sharing platform between neuroscientists and computing researchers. That was great, but CCC was more ambitious than what we were thinking, and said, ‘Go, bring government folks and industry folks in, too,’ and so on.” Indeed, a central goal of CCC’s visioning initiatives is to convene the computing community broadly — incorporating the perspectives of academics, industry representatives, policymakers, and more.

Still, the proposal development process remained a two-way street. Having organized 93 visioning workshops as of Spring 2026, CCC’s mission in evaluating proposals is to maintain the proposers’ intent and goals and to play to their particular strengths while also leveraging CCC’s experience to maximize the initiative’s impact.

“I think the sense we got through the entire review process was that the Council was completely supportive of the general vision and the idea,” Abhishek reflected. “It didn’t feel like an adversarial conference review process. It felt more like, ‘We want to see this succeed, so here are the pieces that we think would help you based on other visioning workshops that we’ve successfully helped shepherd.’”

In total, this process, from initial interest form to final proposal acceptance and the green flag to begin planning, took about three months. 

In the upcoming second part of this series, learn more about the workshop-organizing and report-writing process for this visioning initiative. It’ll explore the process to actually get the Neural Interfaces workshop off the ground and then synthesize the workshop insights to shape the upcoming report.

Stay Tuned for the Full Workshop Report and Get Involved with CCC Visioning

If you’re interested in learning more about how to get involved with CCC visioning initiatives or want to propose one yourself, we encourage you to read more about visioning proposals here. For the full scope of what it’s like to organize a visioning initiative with CCC, you can read our Visioning Best Practices document.

The full workshop report for Defining the Role of Computing Interfaces in Neural Interface Design will be released in Summer 2026, laying out key findings and strategic recommendations for the future of computing in neural interfaces. Subscribe below or follow us on LinkedIn to be the first to know when it’s released.

Tune in to the CCC LinkedIn Showcase Page for updates and more Visioning in Action. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

Advancing Cutting-Edge Technology 

Bringing computational and citizen science/crowdsourcing research together has the potential to reshape our scientific capacity, leading to more accurate and timely discoveries, applications, and monitoring. Already, citizen scientists map biodiversity on iNaturalist, analyze protein folding configurations to advance drug discovery on Foldit, and determine antibiotic resistance through Bash the Bug. These users possess the knowledge, creativity, and skills that make up for what technology cannot do alone and push scientific research forward. 

Convergence offers a unique opportunity to do vital research into building trust, access, and transparency into AI systems by embedding them in real-world participatory contexts. Citizen science projects provide a built-in user base, ultimately making technology like AI safer and more trustworthy through large-scale experimentation. Key to the success of this venture is the breaking down of silos and increased knowledge-sharing between academia and industry — to the benefit of both.

Notably, these contributors can also enrich AI models used in research with contextual insights and provide essential edge-case observations that increase the validity of data, making the most of AI’s strengths while avoiding its downfalls. But their impact has broad applications across industry sectors. Citizen scientist/crowdsourced contributions can improve weather forecasting through on-the-ground reporting, for example; convergence thus opens the door to improving our natural resource management, agriculture, and more. By harnessing the strengths of citizen scientists, crowdsourcers, and computational technologies, we can accelerate the scientific process across disciplines, reliably produce world-leading innovations, and elevate American industry.

Creating a Tech-Minded Populace

Large-scale convergence also promotes the formation of a more technologically-literate society. To make convergence work on the participant side, non-expert users need to understand, in a simpler way, how advanced computing technology like AI works and how to use it. Having such exposure may encourage the use of such technologies in other aspects of life as well as introduce it to brand-new users. Removing the barrier for entry and the legitimizing of these tools through their very visible use in scientific research projects will make their benefits clear to everyday users. 

Read the Full Report

For a full picture of the impact of large-scale convergence, as well as key recommendations across sectors for how to make it a reality, we encourage all members of the computing community to read the Grand Challenges for the Convergence of Computational and Citizen Science Research report below.

Read the Full Report Here

Join the Community Chat

For another way to engage with the findings in this report, join us for an upcoming CCC Community Chat on May 20th, 2026 at 3:30pm. This 45-minute Zoom webinar, hosted by the report’s authors, will dive into key findings, follow up with a series of lightning talks with experts from citizen and computational science, and end with a Q&A. Featuring guest speaker Marc Kuchner, Citizen Science Officer at NASA!

Register and Submit Questions Here

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

The efforts of citizen scientists have enabled research projects across scientific disciplines. From users mapping biodiversity on

Join us for the second-ever CCC Community Chat: A New Era of Scientific Progress: Uniting Advanced Computing and Citizen Science for Accelerated Research. During this 45-minute webinar, the authors of the CCC Grand Challenges for the Convergence of Citizen Science and Computing Research report will address the strategic value of bringing citizen science research and advanced computing together and how the two might mutually enrich each other. This large-scale convergence promises enhanced results for both disciplines and has the potential to fundamentally reshape how we conduct scientific research. In doing so, it could usher in a new era of scientific progress.

Join for a Live Q&A with the Report’s Authors and Guest Marc Kuchner (NASA)

For this special event, we’re excited to announce guest speaker Marc Kuchner, Citizen Science Officer at NASA. He will join the webinar for a short lightning talk on the citizen science projects underway at NASA that are enhancing our understanding of the universe.

Our second lightning talk will be given by Tanya Berger-Wolf (The Ohio State University), also a lead author on the report. Tanya will present about how computing research converges with citizen science research in conservation biology projects.

Hear key findings from the report, including:

• Critical opportunities for human-centered computing research
• The ways technologies like AI and drones can augment citizen science projects
• How convergence is key to addressing pressing national priorities
• What infrastructural changes are needed to make large-scale convergence happen

At the end of the webinar, Marc and Tanya, along with the report’s other lead authors — including Lucy Fortson (University of Minnesota), Lea Shanley (International Computer Science Institute and GNIES, University of Wisconsin-Madison), Kevin Crowston (Syracuse University), Corey Jackson (University of Wisconsin-Madison), Saiph Savage (Northeastern University) — will be happy to answer any questions during a live Q&A. 

Register Now!

Event Date: Wednesday, May 20, 2026

Event Time: 3:30pm ET

We hope you will join us for this amazing opportunity to delve deeper into the future of human-centered computing. Please use the link below to register for the webinar, and feel free to submit some questions ahead of time.

Register Here

Tune in to the CCC LinkedIn Showcase Page for updates and more Community Chats like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.

To address this growing challenge, the U.S. Center for AI Standards and Innovation (CAISI), housed within the National Institute of Standards and Technology (NIST) at the Department of Commerce, released a Request for Information (RFI) on practices and methodologies for measuring and improving the secure development and deployment of AI agent systems. The Computing Community Consortium (CCC) and Computing Research Association (CRA) recently submitted a response to this RFI, providing key insights and recommendations for the future of AI security.

Identifying New Roadblocks

AI agent systems pose unique challenges when it comes to security threats and vulnerabilities. One of the most significant difficulties is precisely what can make them so useful: their ability to adapt and learn. This quality means they are inherently more unpredictable, which can make debugging the system more difficult and can lead to problematic actions created by code errors quickly escalating.

Some of the other questions current security models are grappling with include:

• How to delegate credentials to an AI agent
• How to identify risk levels around highly context-dependent and unpredictable AI agents and their users
• How to determine accountability for security failures
• How to avoid cascade effects when one compromised agent taints the decision-making of downstream agents

The trend towards replacing human workers with AI programs in activities like coding poses its own particular security risk. Humans who understand the underlying systems behind AI agents, how they function, and how they generate their output are essential to tackling emerging threats. To that end, this new RFI recommends that agencies like NIST set expectations around terms like “human-in-the-loop” practices, building on the NIST AI Risk Management Framework and Playbook to provide implementation guidance.

Fostering Security in the Age of AI

Other ways to increase the security of AI agents include fundamentally reevaluating how we approach typical cybersecurity. Some first approaches suggested in this RFI include:

• Examining other periods of major leaps in abstraction in computing for applicable lessons
• Reevaluating the concepts of fuzz testing and input sanitization failures in the dynamic context of autonomous agents
• Looking closely at the extent to which AI agents are capable of policing other AI agents
• Assessing the particular risks of multi-agent systems, especially when the different agents have differing goals
• Investment in research looking for effective guardrails that could be embedded within AI frameworks or could withstand assaults from such systems
• Targeting research efforts on assessing if a human or another AI agent is interacting an AI agent
• Forging stronger connections between the computing research community and industry cybersecurity practitioners

Historically, much of cybersecurity takes place after the design or deployment of an application. This RFI recommends a fundamental shift away from this practice, encouraging NIST to create incentives for industry to make security by design in AI systems rather than a second thought.

Read the Full Response

For the full scope of CCC and CRA findings and analysis on advancing the security of AI, access the full response below. You can also see more CCC responses to the community here.

Read the Full RFI Response Here

Please note any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the authors’ affiliations, or of the National Science Foundation, which funds CCC.

From users mapping biodiversity on iNaturalist to analyzing protein folding configurations to advance drug discovery on Foldit, we have seen the value of engaging the public in scientific research projects. This report also presents the opportunity of citizen science and computing research, together, to maximize the potential of tools with broad applications like artificial intelligence (AI). On a large scale, such convergence is capable of directly addressing some of our most pressing national challenges and priorities.

Supporting National Interests

The impact of the convergence of citizen science with advanced computing aligns directly with a wide array of national priorities:

• It would enhance national competitiveness due to the increased productivity, agility, and reach of scientific disciplines across the board.

• The extensive volunteer efforts involved in convergence would likewise increase economic value and increase government efficiency by reducing the duplication of efforts, streamlining workflows, and lowering the cost of data collection and analysis across state and federal agencies.

• Citizen science offers incredible workforce skills development for those who participate, enhancing scientific literacy among the population as well as serving as accessible entry-points for STEM careers.

• Large-scale convergence requires investment in robust infrastructure that may help address local, state, federal, or even international scientific challenges, including security and privacy concerns.

• An increase in convergence harnesses public data in a way that enhances the credibility of scientific research, accessing perspectives and local knowledge that traditional scientific research methods may overlook as well as emphasizing appropriate data sharing and reuse.

• Participation in citizen science establishes a sense of civic identity and public agency by offering participants a feeling of ownership over their contributions and the eventual results and innovations they enable.
  

• Data collected by citizen scientists is especially valuable in improving natural disaster response, where individuals can provide fast and specific data points about the impacts of events like hurricanes, earthquakes, or tornadoes.

Developing Critical Infrastructure

Large-scale convergence requires investment and coordination to reach its full potential. Workshop attendees crafted a strategic roadmap for the sustained platforms, governance systems, and physical/cyber architecture required to support scalable, trustworthy, and nationwide convergence efforts:

• Encourage and incentivize cross-agency collaboration: Break down silos by integrating citizen science and crowdsourcing with government agency AI, cloud, and technology strategies.

• Create permanent federal funding streams for convergence infrastructure: Ensure stable funding for long-term platform and infrastructure maintenance and evolution.

• Build interoperable and AI-ready data infrastructure for participatory sciences: Define national standards (FAIR data, APIs, metadata) for data exchange across platforms and projects.

• Develop scalable provenance frameworks for participatory AI systems: Fund research on lightweight, privacy-aware provenance capture that spans human actions, AI model evolution, and distributed execution, enabling reproducibility, auditing, and long-term reuse at the national scale.

• Invest in privacy and security-preserving data frameworks: Implement and evaluate mechanisms (e.g., federated learning, data trusts, open standards) that are core to participatory science infrastructure, and formalize the necessary governance and legal structures for secure data use.

• Establish national guidance for explainable, transparent, and trustworthy AI in citizen science: Develop research-backed guidelines for interpretable outputs, community consent, and auditability.

• Develop next-generation participatory AI governance: Co-create frameworks that give participatory science communities real agency over AI deployment decisions, data use, and model evolution.

• Launch a National Citizen Science & AI Convergence Hub: Establish a central, virtual hub to share tools, standards, best practices, case studies, and training materials, thus reducing duplication and accelerating adoption.

Read the Full Report

For a full picture of the impact of large-scale convergence, as well as key recommendations across sectors for how to make it a reality, we encourage you to read the Grand Challenges for the Convergence of Computational and Citizen Science Research report below.

Read the Full Report Here

Join the Community Chat

For another way to engage with the findings in this report, join us for an upcoming CCC Community Chat on May 20th, 2026 at 3:30pm. This 45-minute Zoom webinar, hosted by the report’s authors, will dive into key findings, follow up with a series of lightning talks with experts from citizen and computational science, and end with a Q&A. Key speakers to be announced shortly!

Register and submit your questions here.

Tune in to the CCC LinkedIn Showcase Page for updates and more reports like this. Stay connected with CCC for the latest insights, publications, and opportunities to engage by subscribing here.