Federal agencies overseeing ADA Title II digital accessibility have extended their compliance timelines. The U.S. Department of Justice has moved its deadline to April 24, 2027, and the U.S. Department of Health and Human Services has extended its deadline to May 11, 2027. These changes apply to all university digital content, including instructional, administrative, and public-facing materials, regardless of where or how that content is hosted.

Although the deadlines have shifted, UGA leadership has emphasized that the University’s commitment to digital accessibility remains unchanged. The additional time is intended to help colleges, departments, and administrative units strengthen accessibility practices and complete the review and remediation of existing content.

Within Franklin College, faculty and staff have already made significant progress toward improving accessibility across courses, websites, and documents. Franklin OIT will continue to support that work throughout the extended timeline with a sustained focus on practical, hands-on assistance.

To help maintain momentum, Franklin OIT will continue offering:

• Workshops focused on creating accessible course materials, documents, and web content (launching summer 2026)

• Open office hours and one-on-one consultations for questions and problem-solving

• Online guides, tools, and resources to support both new content creation and the remediation of existing materials

These efforts are designed to help faculty and staff not only meet compliance requirements, but also adopt long-term, sustainable accessibility practices that support inclusive teaching, learning, and operations.

Franklin faculty and staff are encouraged to continue working toward full accessibility well ahead of the new deadlines. Information about upcoming workshops, office hours, and available resources will be shared regularly.

Questions about accessibility support or compliance can be directed to Franklin OIT or to unit-level digital accessibility liaisons.

Franklin College appreciates the continued commitment across the college to improving digital accessibility and ensuring equitable access for all members of the University community.

• accessibility.franklin.uga.edu

• Digital Accessibility Handbook

Franklin College has released new guidance to help faculty and staff understand when and how generative AI tools such as Copilot, ChatGPT, Gemini, and Claude may be used appropriately for coding, scripting, dashboards, and other locally developed solutions.

Based on an April 20, 2026 administrative memo from the Vice President for Information Technology, the guidance reinforces that using generative AI to improve local productivity is encouraged. At the same time, it clarifies that AI‑generated code does not remove existing responsibilities related to security, maintenance, data governance, or long‑term support.

Under UGA’s established Edge–Leverage–Trust technology framework, tools that remain entirely within a unit’s local environment are considered “Edge” activities and are managed at the unit level. Examples include one‑time scripts, Excel macros, local dashboards, and workflow automations that do not integrate with enterprise systems or shared identity services.

In contrast, any tool that connects to enterprise systems such as Banner or PeopleSoft, uses UGA authentication, introduces new data storage, or is intended for use across multiple units is considered a custom application. These tools carry institutional risk and must receive appropriate governance and data steward approvals before development or deployment.

The guidance emphasizes that this distinction applies equally to traditionally written code and AI‑generated code. Units that develop custom applications are fully responsible for development, maintenance, security, user support, and continuity planning. EITS and Franklin OIT do not provide operational support for staff‑built applications that bypass formal governance processes.

EITS and Franklin OIT remain available as consultative partners, particularly when evaluating vendor solutions, assessing security or data implications, or reviewing architectural fit. Early consultation is encouraged when a project may extend beyond local use.

In general, generative AI tools are well suited for drafting, research, data analysis, prototyping, local automation, and other productivity‑enhancing tasks that stay within a unit’s scope. Faculty and staff are encouraged to pause and seek guidance if a tool may affect shared systems, support ongoing operations for others, or create institutional risk.

Additional resources, including UGA’s AI acceptable use guidance and the full administrative memo, are available online. Franklin leaders are encouraged to discuss specific scenarios early to support innovation while protecting the University’s systems and data.

Additional Guidance on AI and Custom Applications and Scripts

The Georgia Advanced Computing Resource Center (GACRC) has released an updated policy clarifying expectations for the use of artificial intelligence (AI) workflows and high-performance computing resources. The changes come in response to a sharp increase in demand for GPU-intensive workloads driven by AI-assisted research tools.

These updates are intended to ensure fair access to shared resources, maintain system performance, and guide research groups in the responsible use of emerging technologies.

GACRC reports a significant rise in the use of AI-enabled workflows across campus, including code assistants, automated job generation systems, and model-serving frameworks. While these tools have expanded research capabilities, they are also placing increased pressure on shared infrastructure, particularly high-end GPUs such as A100 and H100 systems.

To address this, the center has clarified how these resources are to be used going forward.

Under the revised guidance:

• High-end GPUs are reserved for batch workloads. Shared A100 and H100 GPUs are intended primarily for scheduled batch jobs. Long-running or persistent interactive use, especially with low or intermittent utilization, is no longer permitted.

• Interactive sessions must be short. Researchers may still use interactive sessions for development and debugging, but these should be limited to a few hours and not used as ongoing work environments.

• Persistent services are not allowed on shared nodes. This includes model-serving endpoints, long-lived notebooks, AI assistants, and agent-based systems. Research groups needing these capabilities are encouraged to work with GACRC to identify alternative solutions.

• Login nodes are restricted to lightweight tasks. Activities such as file editing, job submission, and monitoring are allowed. Running persistent processes, including remote IDE backends like VS Code or Cursor, is prohibited. Approved tools such as Open OnDemand should be used instead.

• AI workflows must be controlled. Users remain fully responsible for all actions performed under their accounts, including those initiated by AI tools or automated agents.

The update also reinforces existing state-level restrictions. In alignment with the Archnews memo issued March 26, 2025, by USG Chancellor Sonny Perdue, certain software applications are prohibited on state devices, including the DeepSeek AI chatbot.

Principal Investigators (PIs) are expected to ensure compliance across their research teams. Specifically, GACRC emphasizes that groups should:

• Use batch processing for production-level GPU workloads

• Restrict interactive GPU use to short debugging sessions

• Avoid running persistent services on shared infrastructure

• Release GPU resources when not actively in use

• Use approved development environments

• Refrain from installing prohibited software

GACRC will continue active monitoring of system usage to preserve equitable access and performance. The center has identified several patterns of concern, including:

• Long-running interactive sessions on shared GPUs

• Persistent or service-like workloads

• Excessive automated job submissions

• Resource usage that negatively impacts other users

If these issues persist, GACRC may take corrective action. Measures may include stricter usage limits, scheduling policy adjustments, job termination, or suspension of user access on a temporary or permanent basis.

Recognizing that AI workflows are evolving rapidly, GACRC encourages researchers with advanced or non-standard requirements to engage directly with support staff. The center has indicated a willingness to collaborate with PIs to develop sustainable solutions that meet both research and infrastructure needs.

The complete policy statement is available here:
https://wiki.gacrc.uga.edu/wiki/Policies#AI_Workflows_and_GPU_Usage_Expectations

For assistance or questions, researchers may contact GACRC Support.

This policy update reflects a broader effort to balance innovation with shared resource stewardship, ensuring that AI-driven research can continue to scale effectively within a collaborative computing environment.