AI in RA
« AI in RA

#3 — AI Act guidance, enforcement, US frontier model influence

July 8, 2026
New guidance for the EU AI Act details requirements for transparency, high-risk systems, and marking AI-generated content, with some obligations becoming enforceable in August 2026. In the United States, the federal government is moving to influence frontier models before their release, while a complex patchwork of over 40 state-level laws emerges. Concurrently, higher education institutions struggle with AI adoption without clear governance, even as studies show AI's potential to improve administrative efficiency and project management.

AI Regulation and Policy

AI in Research and Project Management

AI in Higher Education Administration

Tools and General Applications

Sources

  1. The EU Digital Rulebook in Motion - Eversheds Sutherland
    EU digital regulation is moving on several fronts, with the AI Act at the centre. Under the Digital Omnibus, one proposal addresses AI and another covers data, privacy and cybersecurity. Alongside that package, the AI Act has its own rollout, including recent guidance on transparency, high-risk classification and AI-generated content marking. Timelines now matter for AI governance, vendor due diligence and product planning. From 2 August 2026, the AI Act's transparency obligations become enforceable. Non-compliance can attract fines of up to €15 million or 3% of global turnover. With changes…
  2. State of AI 2026 June Update: Governments Became Gatekeepers as AI Makes a Political Turn - Serious Insights
    June 2026 confirmed that AI is no longer governed solely by model capabilities, corporate ambition, or user adoption. It is now shaped by political review, infrastructure scarcity, frontier-model access controls, global model competition, memory economics, and the movement of AI from apps into operating environments. The May update argued that capital was concentrating while trust and infrastructure lagged. June added a harder edge: the U.S. government moved from a delayed policy posture to direct pre-release influence over frontier models. President Trump's June 2 executive order, Promoting…
  3. The AI Compliance Patchwork: Navigating State and Federal AI ...
    The EU AI Act's Article 50 transparency requirements take effect on August 2, 2026, while over 40 US states have introduced or enacted differing AI legislation, creating a complex compliance landscape. Federal enforcement is fragmented across agencies like the SEC, FTC, and FDA, which are extending existing authorities to cover AI systems. Key state laws include Texas TRAIGA, California SB 53, and Colorado ADMT Act, each with specific obligations and timelines.
  4. 50,000 Students And Faculty Just Revealed Higher Ed's Top AI ...
    Two large higher education AI surveys involving 50,000 students and faculty globally reveal that AI adoption in higher education is outpacing institutional response, leading to students using AI without guidance and faculty disengaging from teaching with it. This creates a "confidence gap" where faculty believe their curriculum is current in terms of AI skills, but students overwhelmingly disagree and are even considering changing majors due to AI's impact on the job market. The core issue is "adoption without authority," where institutions lack clear governance, preparation, and purpose in…
  5. Artificial Intelligence in large-scale project management: from value management office automation to strategic stewardship
    Large-scale project delivery has always been hard. Data volumes grow, stakeholder networks sprawl across time zones, and the margin for error narrows as program budgets rise. What has changed in the past decade is that artificial intelligence (AI) now offers project organizations a credible set of tools to manage that complexity, not just by automating paperwork, but by reshaping how governance itself is structured. This article examines that reshaping, focusing on the organizational transition from the traditional Project Management Office (PMO) to the emerging Value Management Office (VMO).…
  6. Best AI for Grant Writing in 2026: Win More Funding with ChatGPT in ...
    This guide compares the best AI tools for grant writing, shows you how to use ChatGPT inside Google Docs, and provides useful prompts for grant writers. It addresses common struggles with AI tools in grant writing and highlights GPT Workspace as a practical solution for collaborating on Google Docs. The article outlines a step-by-step process for leveraging AI in different sections of a grant proposal, such as needs statements, project narratives, and budget justifications.
  7. The AI Bid Surge: Navigating the 2026 Grantmaking Revolution
    Nonprofits can survive the 2026 "bid surge" by adopting Augmented Authenticity—using AI for ethical data synthesis and first-draft generation while preserving human-led storytelling. This hybrid strategy bypasses automated foundation filters, reduces administrative burden, and maintains the emotional resonance required to win high-value grants. The "Volume Trap" of generic AI proposals is real, and ethical governance and human-in-the-loop approaches are crucial for success in the 2026 grantmaking landscape.
  8. EMA Research Finds AI-Driven Operations Require an Enterprise Control Plane
    A new study reveals that organizations are cautiously expanding AI-driven operational authority as enterprise control becomes increasingly federated. Enterprise Management Associates (EMA™) released a research report, "From Outcomes to Authority: Defining the Enterprise Control Plane," authored by Dan Twing, which surveyed 336 enterprise IT professionals. The research explores how organizations are adapting operations as AI integrates into automation, orchestration, observability, service management, and cloud platforms. It finds that enterprise operations are increasingly distributed,…
  9. Data Privacy, AI Regulatory, and Compliance Update: June 2026
    June 2026 continued the rapid expansion of privacy and AI regulation, with new U.S. state comprehensive consumer privacy laws, a broad Connecticut AI and online safety statue, and additional implementation activity under existing international AI and privacy frameworks. For businesses, the through-line is operational readiness and regulator expectations: regulators are moving from high-level policy statements toward concrete transparency, governance, contracting, and disclosure requirements. Key takeaways include new comprehensive privacy laws in Vermont and Louisiana, amendments to…
  10. ARTIFICIAL INTELLIGENCE APPLICATIONS FOR OPTIMISING ACADEMIC AND NON-ACADEMIC ADMINISTRATIVE PROCESSES IN SOUTH-SOUTH NIGERIAN UNIVERSITIES
    This study investigated the effect of Artificial Intelligence (AI) Applications for Optimising Academic and Non-Academic Administrative Processes in South-South Nigerian Universities. Anchored on Fred Davis’s (1986) Technology Acceptance Model (TAM), the study examined how AI tools influence process efficiency, data accuracy, and decision-making within higher education administration. A descriptive survey research design was adopted, involving twelve universities two (one federal and one state) selected from each of the six South-South states. Using Yamane’s (1967) formula, a sample of 420…
  11. ARTIFICIAL INTELLIGENCE APPLICATIONS FOR OPTIMISING ACADEMIC AND NON-ACADEMIC ADMINISTRATIVE PROCESSES IN SOUTH-SOUTH NIGERIAN UNIVERSITIES
    This study investigated the effect of Artificial Intelligence (AI) Applications for Optimising Academic and Non-Academic Administrative Processes in South-South Nigerian Universities. Anchored on Fred Davis’s (1986) Technology Acceptance Model (TAM), the study examined how AI tools influence process efficiency, data accuracy, and decision-making within higher education administration. A descriptive survey research design was adopted, involving twelve universities two (one federal and one state) selected from each of the six South-South states. Using Yamane’s (1967) formula, a sample of 420…

Also this week

Full transcript
In August 2026, transparency rules from the EU's AI Act become enforceable. That development is one of several we're tracking this week on AI in RA. We begin with the new guidance from the European Union. It looks like the regulatory picture for AI is starting to come into focus, especially in Europe. With the AI Act moving into its implementation phase. The European Commission has released guidance on transparency, classifying high-risk systems, and how to mark AI-generated content. And there's a hard deadline on the calendar: August 2, 2026. That's when the transparency rules become enforceable. Enforceable with serious penalties. The fines can go up to 15 million euros or three percent of global turnover. That gets a company's attention. It does. And that approach contrasts with what's happening in the United States, where the federal government seems to be shifting its strategy. Right, they're moving more toward direct, pre-release influence over the most advanced AI models. It's less about setting broad rules after the fact and more about engaging with developers before a model goes public. We're already seeing that play out. OpenAI's limited preview for its GPT-5.6 models—Sol, Terra, and Luna—came after they engaged with government entities. And it also explains what happened with Anthropic. Their planned rollout for the Fable 5 and Mythos 5 models got tied up in federal security reviews. So the federal government is establishing its lane. But then you have the states, which are creating this complex compliance environment. It's a patchwork. More than 40 states have introduced or enacted their own AI laws. So a company has to navigate not just federal policy, but a web of different state-level obligations. Like Texas's TRAIGA law, California's SB 53, Colorado's ADMT Act… the list keeps growing. And that's on top of federal agencies like the SEC and FTC extending their own existing authority to cover AI. So that's the legal and regulatory framework. But what about how organizations are actually integrating this technology? There's a report on higher education that found a problem described as 'adoption without authority.' Meaning institutions are just bringing AI in without any real governance, preparation, or even a defined purpose for it. It's creating confusion. And a 'confidence gap.' The report says faculty members believe their curriculum is up-to-date on AI skills, but students don't agree. They're the ones who see the gap. Though in some places, AI is showing clear benefits. A study in South-South Nigerian universities found that AI applications improved administrative efficiency, especially in data management. But there was a condition. Let me guess: infrastructure. The study noted that federal universities saw higher benefits because they had more robust digital systems to begin with. Exactly. And we're seeing a similar story of measurable improvement in the corporate world, specifically in project governance. Machine learning is being used to improve estimation accuracy and risk detection. And the numbers are significant. One finding suggests deep learning models can hit 85 to 90 percent accuracy in project cost estimation. That’s not a minor tweak; that changes how you do business. It leads to organizations giving more operational authority to AI systems, especially as their control structures become more distributed and federated. This is all playing out in very specific domains, too. Look at grant writing. Right. New guides are comparing different AI tools for the grant writing process. One tool, GPT Workspace, is designed for collaborative AI grant writing. And a new strategy is emerging for nonprofits called 'Augmented Authenticity.' The idea is to use AI for the data synthesis and first drafts, but keep a human-led narrative to manage the expected increase in AI-assisted applications. It's about maintaining a human touch while using the tools for efficiency. There are also a few other developments on the radar. Quickly, the EU Council adopted some amendments but delayed enforcement for the high-risk rules in the AI Act. An enterprise survey found that while two-thirds of workers think AI improves efficiency, they're divided on what it means for job security. And on the state level, Vermont and Louisiana have new comprehensive consumer privacy laws, while Delaware amended its existing one. The regulatory landscape continues to evolve. That covers the latest developments for now. We will continue to track these stories and more. From AI in RA, thanks for listening.

From our community

Builders Group

RSVP →

From our community

Join AI in Research Administration

Join the community →