This year marks the 10th anniversary of the State of Open Data, a collaboration between Digital Science, Figshare, and Springer Nature, that has tracked researcher attitudes toward data sharing across 192 countries every year since 2016. The 2025 edition draws on a full decade of longitudinal evidence to identify where real progress has been made, where momentum has stalled, and what new forces are reshaping the open data landscape.
When we covered the 2021 and 2023 editions of this report, two themes were already firmly established: growing but uneven awareness of the FAIR data principles, and a persistent gap between researchers' data-sharing efforts and the professional recognition they receive. Both continue in 2025. This year, however, there is also a genuinely new story: the rapid integration of AI tools into research data workflows. Together, the findings portray a field transformed in awareness and culture, yet still constrained by structural barriers that have proven slow to shift.
FAIR Awareness Has Surged
The clearest success of the past decade is the dramatic rise in awareness of the FAIR data principles: Findable, Accessible, Interoperable, and Reusable. When first asked in 2018, nearly 60% of respondents had never heard of FAIR. By 2025, that proportion has fallen to just 20.4%, while active familiarity has nearly tripled from 15.2% to 40.6%.
Support for Open Practices Remains Strong
FAIR awareness is part of a broader picture of durable enthusiasm for open science. Researchers continue to express strong backing for open practices across the board.
Enthusiasm for Open Data Mandates Is Falling Globally
One of the more consistent longitudinal findings is the near-universal decline in strong support for national open data mandates across all ten countries tracked since 2016. While researchers broadly endorse openness as a value, strong support for formal mandates has softened over the same period.
Researcher “Strong Support” for a National Open Data Mandate, 2016–2025
% of researchers responding “strongly support” when asked about a national mandate for making research data openly available. Shown at selected intervals. Sample sizes range from n = 945 (2016) to n = 2,410 (2025); total longitudinal sample N = 21,555.
Declines are steepest in Australia (63.2% to 27.4%) and the United States (52.9% to 29.7%). By contrast, India remains the most supportive country, declining only modestly to 54.7%. China sits lowest at 25.8%.
What the data makes clear is that blanket mandate strategies are unlikely to work. Pairing policy with practical infrastructure, training, and institutional support is essential to sustaining researcher buy-in over the long term.
The Credit Gap: Still the Most Persistent Barrier
If there is one finding that has shadowed all ten years of this report, it is the "credit gap" - the widespread perception that professional recognition for data sharing is fundamentally misaligned with the effort it demands.
Researchers who say they receive too little credit (%)
Q: “Do you think researchers currently get sufficient credit for sharing data?” Longitudinal data from six annual surveys (N = 28,584)
After five years of growing awareness and expanding policy attention, the overwhelming majority still perceive a fundamental gap between the work of sharing data and the professional rewards it yields. Until data sharing carries tangible weight in hiring, promotion, and funding decisions, this structural disincentive will continue to undermine the broader momentum the field is building.
AI Tools Are Rapidly Entering Research Data Workflows
The most consequential new development in this year's report is the striking acceleration of AI tool adoption across research data workflows. The survey began tracking this in 2024. In just one year, the picture has transformed.
Active use of AI for data processing jumped from 22.1% to 31.9%, driven largely by a surge in researchers who recently started using these tools (14.4% to 21.7%). The pool of those "not aware" or "aware but not considering" contracted by about 9 percentage points, signaling that hesitation is rapidly diminishing. The same momentum is visible in metadata creation (16.1% to 25.1%) and data collection (17.2% to 23.5%).
The potential upside is significant. AI could meaningfully support FAIR-aligned practices, reducing the overhead of metadata creation and format standardization. But as adoption accelerates, ensuring transparency, reproducibility, and responsible use becomes critical to guaranteeing that AI-assisted research data remains trustworthy and genuinely reusable.
Summary: Three Cross-Cutting Priorities
The data reveals three critical insights about the current state of open science:
Knowledge Without Support
Awareness of open science principles has grown, but lack of credit and resources still hinder implementation. Mandates alone are not enough and practical support is essential.
Regional Divergence
Support for data mandates varies widely, showing that one-size-fits-all strategies are ineffective; regional approaches that account for local research cultures and infrastructures could work better.
Disciplinary Disparities
STEM fields lead in FAIR adoption while the Humanities and Social Sciences lag, underscoring the need for discipline-specific support and training.
The full 2025 report and interactive survey dashboard are available at stateofopendata.com.
