
Most UK companies are already using AI. Sales teams draft proposals with it, operations teams summarise contracts, and finance teams work through spreadsheets. The tool people settle on is often Claude, because it copes with long documents, data analysis and the routine workflows they would rather hand off.
Nor is it only junior staff: TrustedTech research in May found that 62% of UK senior leaders use AI tools their organisation has not approved. That leaves many companies with a question they cannot yet answer: where does the data go when staff use AI? Under UK GDPR, a prompt containing a customer’s details is personal data.
Who will ask
Sooner or later, someone will ask: an enterprise customer’s security questionnaire, an investor’s due diligence, an insurer at renewal or a regulator. The second question, usually from legal, is what happens if the AI supplier stops trading.
Why banning does not work
The instinctive response is to block public AI tools. It rarely lasts more than a month. The use moves to personal phones, and the company loses the little visibility it had. Most companies end up asking one question in different words: how do we keep the Claude our people already like, with the security we need?
The better route is an approved tool that is genuinely better than the free one: it sees the company’s own documents and systems, so people stop pasting them in, and it runs somewhere the company controls.
What secure AI looks like
A company should be able to describe its AI use in four sentences:
- It runs inside the company’s own cloud, in a UK region, so data does not leave the business and is not used to train anyone’s model.
- Access follows the permissions already in place, so each person’s assistant sees what that person can see and nothing more.
- Every request is logged somewhere the supplier cannot edit, and a named person signs off anything that reaches a customer.
- If the supplier disappears tomorrow, the tool keeps running, because the environment, the keys and the logs belonged to the company all along.
The model itself does none of the second and third points. They come from the management layer around it: who can install what, which systems it connects to, what gets logged.
That is the layer Moterra deploys, inside a company’s own AWS or Azure account. The solution is certified end-to-end to ISO 27001, ISO 42001 and SOC 2 Type II, and is defensible under UK and EU GDPR and DORA.
What comes next
The security question is the urgent one. The bigger change is quieter. A growing number of people now open Claude first thing in the morning and leave it open all day. They tell the assistant what they need, and it reads the inbox, pulls the customer record, drafts the reply and updates the CRM. The person’s job becomes checking, deciding and signing off.
Nobody can say yet how far this goes. What is already clear is that an assistant which acts across every system is only as safe as the place it runs. Companies that set it up in their own cloud now can add each new step as it arrives. Those whose staff still use it on personal phones and laptops will have to start again.
“Companies think they have to choose between moving fast with AI and keeping control of it,” says Vytenis Sakalas, CEO and co-founder of Moterra. “They don’t. That choice disappears once AI runs in your own cloud.”
So, can AI be secure?
Fifteen years ago, boards asked the same question about the cloud. Most held back, a few moved early, and within a few years running sensitive work in AWS or Azure became normal rather than risky. AI is following the same path, only faster. The companies that treat it as part of their own infrastructure today will look back on this question the way they now look back on the cloud debate.
Moterra is a private AI deployment company and a Select Services Partner in the Claude Partner Network. We help businesses deploy Claude inside their own AWS or Azure environment, connect it to the tools they already use, and keep it secure and governed, so everyday processes get faster.