The numbers from the 2026 State of the Database Landscape: AI Edition are striking. 99% percent of respondents using AI report at least one measurable benefit for their database work. Automation is up, performance is improving, and three-quarters report significant cost savings. By almost any measure, AI is delivering. However, sitting alongside that near-universal positivity in the same dataset, security and privacy concerns have climbed to 64%. Regulatory compliance anxiety has risen to 40%. Nearly two-fifths of organizations are still relying on manual approaches to test and deploy database changes, and only 23% have a formal data governance framework in place.

And, if that wasn’t enough, more than half are knowingly accepting higher security risk as a direct trade-off for the efficiency gains that AI provides.

These are patterns worth understanding for those responsible for the teams, systems, and practices that AI is now running through at pace.

How AI in database management is delivering real, measurable value

AI adoption in database management is significant, with usage nearly tripling year-on-year (from 15% to 44%). This is driven primarily by large, cloud-based, multi-platform environments where operational complexity is at its highest, offset by the potential gains from automation being at its greatest.

This is far from just a cautious experimentation around the edges. In fact, for a growing number of teams, AI is now woven into how work gets done.

AI is being utilized for a wide range of tasks: data quality assurance, schema design, query optimization, code review, and automating database management, to name a few. Broadly speaking, with its implementation across the entire development cycle, AI’s adoption is genuinely meaningful.

Interestingly, the most significant shift has been in the use of AI for data quality work. This has more than doubled since last year, reflecting a broader shift in how teams are thinking about what AI is actually for. Rather than simply using it to write code faster, it’s increasingly being used to manage and validate the data those systems depend on. And, while the reported benefits aren’t groundbreaking, they clearly are having a positive impact. Task automation, for example, is cited as a benefit by as many as 63% of respondents – followed by improved database performance (60%), and efficiency gains (59%).

Perhaps most interesting is the jump in those who view enhanced security as a benefit of AI, up from 25% to 53% year-on-year - the largest single increase of any category. Financial returns are also tangible, with 76% reporting significant cost savings.

For managers and team leads who have been making the case for AI investment internally, these are compelling numbers.

“The places I’m genuinely seeing AI help today are the parts of database work that usually take a lot of time and troubleshooting or debugging. It’s useful for getting to a solid first draft quickly, for translating noisy outputs into something understandable, and for giving engineers a structured starting point during troubleshooting.”

Advait Patel, Senior Site Reliability Engineer, Broadcom

Investment levels reflect this confidence. Around half of organizations are spending more than $100,000 on AI annually, with almost a quarter of large enterprises exceeding $1 million. That’s significant budget being committed, suggesting that - for many organizations - the question has shifted from whether to adopt AI, to how to adopt it well.

Why AI adoption is outpacing database governance

So, why is the rising adoption of AI increasing security risk? Well, AI is accelerating changes across environments that were already operating with inconsistent foundations – in turn, amplifying the gaps that were already there, rather than creating ones that didn’t previously exist.

Take testing and deployment as a concrete example. Despite the pace of AI-driven development, 39% of organizations are still relying on manual processes to test and deploy database changes, and these don’t scale in proportion with the rate of change AI enables. Instead, they become a bottleneck and - more critically - a source of inconsistency.

Take changes that move faster than the ability to validate and trace these errors, for example. These errors don’t simply disappear – they just travel further before they’re caught, and the cost of catching them increases.

Governance tells an equally revealing story. Only 23% of organizations use a formal data governance framework, yet 61% have undergone a compliance audit in the last twelve months. That’s a significant and growing gap between the scrutiny organizations are subject to and the structures they’ve built to meet it.

Furthermore, AI is increasing the pace of schema change and data reuse (both of which make traceability, clear ownership, and access controls more important). Where those mechanisms are weak or absent, the consequences don’t tend to announce themselves cleanly - instead, they surface gradually.

“AI can be the good and the bad here. It can analyse threats and find issues automatically so it can strengthen my own system’s security. The concern is mainly how people deal with it or when you give it too much power. Granting AI the permissions to automatically fix everything means it can potentially also cause huge damage.”

Ben Weissman, CEO, Solisyon GmbH

There’s also the compounding effect of environmental complexity to consider. The organizations using AI most heavily tend to be the same ones running the most complex estates.

For example, with multi-platform, hybrid cloud, and on-premises setups, data moves through four or more transformation stages before it reaches the people who depend on it. In these environments, ownership is often unclear, and a control that functions well in one part of the estate may not exist in another.

AI doesn’t create that fragmentation but does expose it, raising the stakes for addressing it in a way that’s hard to ignore for long.

So, can you trust AI with your database?

Perhaps the most telling finding in the report is that 58% of organizations are explicitly accepting higher security risk as a trade-off for AI efficiency gains. It’s tempting to read that as recklessness but, in most cases, it’s a pragmatic calculation made by people who understand the pressures they’re operating under. They’ve simply made a conscious choice about where to place the risk.

The more difficult question is whether those individuals are the same ones who will be managing the consequences when that deferred risk eventually surfaces (and in engineering and DevOps environments, it usually does.)

The trust dimension runs deeper than risk tolerance, though. There’s a meaningful and often underappreciated difference between AI delivering genuine value in a controlled, well-understood environment, and AI being relied upon in organizations where the underlying data quality and governance foundations aren’t strong enough to support it safely.

Gartner’s projection that 60% of AI projects without AI-ready data will be abandoned by the end of 2026 is worth pausing on. The failure mode in most of those cases won’t be the AI itself - it’ll be what the AI was working with.

“There’s still a trust issue with AI… It’s just as confident even when it’s completely wrong.”

Deborah Melkin, Financial Services

For managers and senior engineers, this is where the day-to-day tension tends to concentrate. AI tools are genuinely useful, teams are depending on them, and delivery expectations have been recalibrated to reflect that. But the confidence of an AI’s output is only ever as good as the quality of what’s feeding into it — and in fragmented, hybrid environments, that quality is frequently harder to verify than the output itself suggests. The tool presents certainty; the foundations beneath it don’t always warrant it.

How AI is reshaping database team structures faster than most organizations planned

There’s one further dimension to the risk picture that deserves direct attention, and it’s the one that’s perhaps hardest to quantify in a survey. AI is changing how work is distributed across teams faster than organizations are updating how they’re structured to do that work. Nearly half of respondents — 49% — report a disconnect between their formal job description and what they really do day-to-day.

This is not a minor administrative lag - it’s a signal that responsibilities are shifting in ways that haven’t yet been formally acknowledged, which creates its own set of accountability and ownership gaps.

Hiring patterns are shifting, too, and in ways that carry longer-term implications. Almost half of organizations (49%) report hiring fewer junior or entry-level staff as a direct result of AI adoption. On one level, that reflects real productivity gains — if AI is handling work that would previously have gone to a junior engineer, it’s rational to recalibrate headcount accordingly.

On another level, it raises a less comfortable question about where the next generation of experienced practitioners is going to come from, and what happens to institutional knowledge in environments that are simultaneously getting more complex and less staffed at the entry level.

“AI is already reshaping entry-level tech in ways most organizations haven’t planned for. The real question isn’t just about roles changing but what the bridge looks like when experienced professionals move on and there are fewer junior people coming through to fill those gaps.”

Kellyn Gorman, Advocate, Redgate Software

Organizations are clearly trying to respond. 76% now provide formal guidance on AI use, up from 52% last year - a meaningful increase in institutional attention. But guidance and readiness aren’t the same thing, and the gap between them is where a lot of the day-to-day friction lives. As Grant Fritchey, Advocate at Redgate, puts it:

“The one thing you can absolutely say with certainty when it comes to AI, we don’t know exactly how things are going to change in the longer term — and the fact that so many say they’re working outside their job description backs that up.”

Grant Fritchey, Advocate, Redgate Software

How to build the right foundations for AI in database management

The instinct when reading data like this is to reach for a prescriptive list, but the more useful starting point is probably a single honest question: is the pace of AI adoption in your environment outrunning the practices needed to sustain it safely? For most organizations, the candid answer is yes — at least in some parts of the estate, and often in the parts that are moving fastest.

The survey data shows a consistent pattern: the areas of highest AI activity tend to be the same areas where governance is weakest, where testing is most manual, and where data ownership is least clearly defined. That correlation isn’t accidental. AI tends to be adopted fastest where it delivers the most obvious value, and those are typically the places that were already moving at pace and had the least bandwidth to invest in structural discipline alongside the delivery work.

The window to address this is still open. More than 80% of organizations expect to adopt additional AI tools in the next one to two years, which means there is still time to build the foundations before the next wave lands on top of the existing gaps.

Formalizing governance frameworks, automating change validation, investing in synthetic test data so teams can validate realistically without increasing data exposure — none of these are particularly glamorous priorities, but they’re the ones that determine whether the benefits being reported today translate into something sustainable, or whether the deferred risk accumulates quietly until it doesn’t.

As Jeff Foster, Director of Technology and Innovation at Redgate, puts it:

“Everyone wants to move faster with AI, but few are truly ready for it. It isn’t just about algorithms — it’s whether your data, systems, and teams are prepared to support intelligent automation safely and effectively.”

Jeff Foster, Director of Technology & Innovation, Redgate Software

The question organizations should be asking now is whether the foundations are being improved at the same rate as the adoption.

Read the full findings in the 2026 State of the Database Landscape: AI Edition — including data on AI usage by task, the governance gap, risk trade-offs, and how organizations are adapting.

This post draws on findings from the 2026 State of the Database Landscape: AI Edition, based on a survey of 2,150 IT professionals globally.