Employees are increasingly signing up for SaaS tools that their companies have not authorised, leaving bosses unsure of how to address the issue.
The growing use of shadow IT – where unsanctioned technologies appear within the workplace – is unavoidable, but organisations can adapt.
“It’s happening, and you cannot run away from it,” says David Parry-Jones, CRO at global language AI company DeepL. “Companies can shut down access to certain websites and tools, but employees will just find an alternative.”
The potential dangers include the unintentional loss of confidential information and other data from within the business and, when it comes to shadow AI, there is also the risk that inaccurate information will be generated. According to 2024 research by Microsoft and Linkedin, 78% of AI users are already bringing their own tools to work.
However, one positive impact of shadow IT is that CIOs gain insight into next-generation technology that may make their organisations more productive and profitable. This is especially true with emerging technologies such as AI, where employees’ use of off-the-shelf tools can help business leaders discover viable use cases.
DeepL recently commissioned a study with Forrester Consulting, which underscores why leaders should invest in reliable and specialised AI solutions to ensure consistency, security and accuracy.
For example, organisations that rely on translated materials to improve communication between employees, clients and customers may inadvertently damage their reputation if employees use inconsistent and inaccurate translation tools.
By using specialised AI tools, organisations can exercise greater control over outputs and achieve better results. With DeepL’s translation software, for example, businesses can enjoy a 90% reduction in internal document translation time, and a higher level of accuracy and personalisation.
Shadow AI is adding to concerns because businesses worry about what third parties might do with their data
Business leaders also need to ensure that the tools their employees use comply with regulations in different markets, such as GDPR in Europe.
There remains some apprehension around AI because the technology is disrupting traditional ways of working. These worries are understandable, especially in sensitive sectors such as law, health or the public sector, where a data breach can have dire consequences.
“We saw the same concerns with cloud computing, but attitudes change over time because the benefits ultimately outweigh the risks,” says Parry-Jones. “Shadow AI is adding to concerns because businesses worry about what third parties might do with their data. We have to work hard to build trust. DeepL Pro, which is designed for enterprises and individuals with regular translation needs, does not store any data inputted for translation.”
He adds that education is key to conveying the benefits of using specialised translation tools.
“Internal productivity improves because people in different locations around the world can collaborate and communicate more effectively, which saves a lot of time. Previously, documents had to be sent to an external agency to get a good quality translation.”
Specialised AI tools are also helping businesses in predominantly monolingual markets, such as the US and the UK, where employees must communicate professionally in the languages of local customers in different countries. There is evidence that this can deliver a considerable competitive advantage, especially for SMEs.
As trust builds around the business benefits of AI, organisations may become more willing to share their data. This will be essential if the next generation of AI tools is to provide even more accurate and personalised solutions.
“Business leaders need to understand that to get really customised AI information, whether for translation or something else, they need to give the engine some data to allow customisation. This will happen over time.”
DeepL is already working with businesses to improve the translation of large files (such as multi-page PDFs) that might contain technical terms or important product information, where accuracy is crucial.
In September 2024, the company enhanced its DeepL glossary functionality, allowing companies to translate specific terms and phrases unique to their industries. This improves global brand consistency by accounting for nuanced translations. Among the languages added to the tool are Korean, Danish, Swedish, Norwegian and Romanian.
“This is about using AI to give an industry context to a translation. An organisation defines which words used internally need to be translated in a particular way,” says Parry-Jones. “If everyone is using the same sanctioned AI tool rather than their own, you get more accuracy and personalisation and avoid potential data protection issues.”
Parry-Jones points out that people will always have a role, especially in sensitive sectors like law. In the legal sector, AI is used heavily for translations, but a human still runs his or her eyes over the final documents.
In other sectors, such as customer service, where online chat tools are common, the current accuracy of AI translation is arguably sufficient, Parry-Jones says. This is especially true when there is the option for someone to speak to a human if they need to.
“The best AI is human, and it is used by people to communicate with other people,” he says. “We are often asked questions about the future and whether, for example, the need to learn a language is dead. In many ways, AI translation tools that are accurate and trusted help people learn languages by accelerating their learning. This is good for business people who travel.”
At a time when businesses face numerous challenges, AI can be a money and time-saving enabler. It can help organisations survive economic downturns, become more corporately responsible and remain innovative and relevant.
The challenge over the next year will be to build trust in, and knowledge of, specialised tools to minimise any threats from employees’ shadow AI.
For more information, visit deepl.com
