- Google unveils new AI initiatives, including expanding UK data residency for Agentspace and AI upskilling projects
- Chirp 3 will also be coming to Vertex AI
- Follows Google’s $1 billion commitment to building a new data center in 2025
Google has announced the expansion of UK data residency for Agentspace, offering extra security by allowing British businesses to store their information locally.
Speaking at the launch in its DeepMind’s London HQ, Google Cloud CEO Thomas Kurian outlined the opportunities Google’s UK data residency extension brings for organisations that use Agentspace.
“It’s an important element to give companies confidence that they can use our systems and their data will be private to their own use and that they have full control to keep the data where they need it,” he noted.
Agentspace inclusion
The existing data residency programme in the UK gives Google Cloud users and customers the ability to conduct machine learning processing for Gemini 1.5 Flash, but Google has strengthened its commitment to UK AI by expanding this to include Agentspace.
“Recognizing the importance of the United Kingdom, we already have a cloud region here,” Kurian added. “We’re expanding our program with a new large scale data center, and we host these solutions and models here in the UK so that you can keep your data in the country and meet regulations to train and serve models right here.”
Agentspace is a business-specific AI agent platform, which combines Gemini AI, Google search, and company data all in one place, using emails, documents, tables, and even third party services like Confluence and ServiceNow within a multi-model search agent, designed to help answer complex questions and boost productivity.
That’s why its data residency is “super important”, said BT Group Chief Executive, Allison Kirkby, praising the move for offering “a bit more sovereignty and protection” for UK users.
“Because our overall mission as a company is to be the most trusted connector of people, business, and society. And considering the world we live in, certainly have customers that want to ensure that their data is safe and secure and it’s on platforms that are encrypted in the right way with the right people handling it”
Skills investments
The news follows Google’s $1 billion commitment to building a new data centre, set to open later in 2025.
This comes alongside significant investment into skills and training initiatives, and a Google Cloud Startup Program, offering up to £280,000 for AI startups across the country – all part of a Google’s reinforced commitment to the UK market.
Google is looking to expand AI skill initiatives in the UK, having trained over a million students, educators, and developers in the last ten years. This includes 35 free Skills Boost credits per month for the Google Cloud Skills Boost programme, with a focus on generative AI in the workplace.
When polled, over half of TechRadar Pro readers (54%) said they use AI daily in their work, with a further 22% saying they use AI tools at least once or twice a week. This outlines just how crucial AI safety and skills training is, especially given only 17% of respondents told us they never use AI at work. If you want to take a look at our polls, or participate in the next one, you can do so by signing up to our Whatsapp here.
“We’re investing heavily in startups,” said Kurian. ”For example, smaller companies who use models to run the business. There’s a whole range of these that we’re doing. We have programs at universities to teach computer science programs on how to use models as part of their work. There’s a broad range of things.”
Over 60% of startups in the UK are already working with Google Cloud, and the “Gemini for UK” event confirmed cloud credits for UK startups, alongside technical resources and training. These startups will also gain a community of Google Cloud experts, as well as investors, partners, and other startups.
Chirp 3
Also announced at Google’s event was the introduction of Chirp 3 on Vertex AI – an audio generation model, available in 31 different languages. This ‘groundbreaking’ model will offer 248 distinct voices and eight speaker options, with detailed speech functionality that “captures the nuances of human intonation.”
Whilst impressive, this does raise some questions about its potential for harm. Chirp3 does currently come with significant restrictions in order to mitigate the risk of abuse, but Google is, “working through some of those things with [our] safety team,” Kurian confirms.
Deepfake AI calls have been on the rise thanks to the advancement of similar technologies, and although AI is frequently used in threat detection, accessible tools like these could lower the barriers for cybercriminals looking to impersonate trusted figures to trick victims into handing over personal or financial information.
Researchers recently hijacked AI agents into writing and sending phishing emails for them, proving that AI agents and tools can be a useful accessory for cyberattackers – so Google will have to ensure strict guardrails and safety controls are in place.
Leave a comment