Unapproved AI tools widely used by data professionals

Four in ten data professionals use unapproved artificial intelligence tools at work, raising concerns over security and privacy practices inside data teams, according to new research from tech and AI training provider General Assembly.

The survey of 269 data analysts, database administrators, data scientists and data engineers in the United States and United Kingdom found that 79% use AI in their roles. Of those using AI, 40% said they use tools that their organisation has not approved. More than one in six respondents, 17%, said they primarily rely on free, publicly available AI tools.

General Assembly, which is part of talent solutions group LHH, said the findings highlight a gap between rapid AI adoption and formal training for staff who work with sensitive data.

"It should be deeply concerning to business leaders to see that those who are handling potentially sensitive or confidential data are using AI in such a risky manner," said Daniele Grassi, CEO of General Assembly. "This is what happens when technology adoption outpaces training and skills. It's a clear signal that enterprises face an urgent need to upskill all employees on the proper, compliant and strategic use of AI."

Training gap

The research found that 75% of data professionals have attended some form of AI training. Only 19% said that training focused on their specific role, such as data analysis, engineering or database administration.

Respondents who received role-specific training reported better use of AI at work and stronger results. They were more likely to say they use AI in their day-to-day activities. They were also more likely to say they always or almost always complete their planned daily tasks.

Those with role-specific training reported higher confidence in their AI skills. They were more likely to say that AI had improved data quality on their projects. They were also more likely to say AI had unlocked greater analytical creativity within their teams.

Data professionals with tailored training were more likely to report that AI had enhanced data literacy across their organisation. They also reported that AI made it easier to communicate with business stakeholders about data and insights. Many in this group said they had gained more free time for strategic work as routine tasks shifted to AI tools.

Mixed impact

General Assembly's survey showed that most data professionals see benefits from AI in the workplace. Respondents cited higher productivity, better data quality and increased analytical creativity as positive outcomes from AI usage.

The impact is not universal. Almost a third of respondents, 32%, said they were only "somewhat confident" that they have the skills and knowledge to use AI in their job. A further 5% said they lack confidence in their AI skills.

Not all respondents reported gains from AI. The survey found that 18% said AI had no impact on their day-to-day work. Another 19% said AI had created more work for them.

Among data professionals who do not currently use AI, 38% cited a lack of training as the reason. Over half of non-users, 51%, said they would need clear examples of how AI could support their daily responsibilities before they would start using the technology.

Role-specific focus

The findings suggest structured, role-based AI training may influence both adoption rates and business outcomes. The data indicates that general awareness courses have reached a majority of data professionals. It also shows that more targeted instruction remains rare.

Grassi said the divergence between generic and tailored programmes is becoming more apparent inside organisations.

"Generic AI training has quickly become outdated," said Grassi. "Role-specific training outperforms one-size-fits-all approaches when it comes to driving adoption, ensuring compliant usage, and delivering business outcomes."

General Assembly conducted the survey over a two-week period and gathered responses from data professionals working across sectors in the US and UK. The company said it plans further research into how different forms of AI education influence workplace usage and organisational policy.