From March 28, a change to a long-standing Amazon Echo privacy feature will mean that every Alexa request will be transmitted to Amazon’s cloud by default, rather than being processed locally on the device.

What’s Changing?

For years, Amazon Echo users had the option to keep some of their voice interactions with Alexa private. For example, select Echo devices, including the Echo Dot (4th Gen), Echo Show 10, and Echo Show 15, offered a setting called “Do Not Send Voice Recordings.” This allowed Alexa to process certain requests locally on the device, meaning users’ voices never left their homes. However, this is about to change.

Every Command To Be Sent To Amazon’s Servers

Amazon has confirmed that from March 28, 2025, every single voice command issued to an Echo device will be transmitted to its cloud servers for processing. The company is discontinuing the local processing option entirely, regardless of user preference. Even if customers select the “Don’t save recordings” setting, voice data will still be sent to Amazon’s servers, although the company says it will delete it once processed.

Emailed Explanation

In an email to affected users, Amazon explained the change, stating:
“We are reaching out to let you know that the Alexa feature ‘Do Not Send Voice Recordings’ that you enabled on your supported Echo device(s) will no longer be available beginning March 28, 2025. As we continue to expand Alexa’s capabilities with generative AI features that rely on the processing power of Amazon’s secure cloud, we have decided to no longer support this feature.”

Why Is Amazon Making This Change?

In short, the official reason for the change is generative AI. Amazon says Alexa is evolving, and the latest advancements require more computational power than what’s possible on an Echo device. Instead of processing requests locally, Alexa will therefore leverage Amazon’s cloud-based AI to provide more sophisticated responses.

This change aligns with Amazon’s push for Alexa+, a new AI-powered version of the voice assistant expected to roll out later this year. The upgrade promises a more human-like, conversational Alexa experience (closer to ChatGPT), offering smarter interactions and better understanding of user requests.

According to Panos Panay, Amazon’s Senior VP of Devices & Services:
“Alexa+ is more conversational, smarter, and personalised. It understands what you mean, even if your request is half-formed or vague. This kind of AI capability requires cloud processing.”

What Does This Mean for Echo Users?

For those who had privacy concerns and enabled “Do Not Send Voice Recordings,” this update removes an important layer of control. Essentially, from March 28 onwards:

– Everything you say to Alexa will be sent to Amazon’s servers.

– Local processing will no longer be an option.

– If you select “Don’t save recordings,” it won’t store your voice data, but Amazon will still process it in the cloud.

– Voice ID (used for personalised features like recognising different users) will no longer work unless you allow recordings to be saved.

– For users who bought their Echo devices under the assumption that they could keep their voice data private, this could feel like a fundamental change to what they originally purchased.

How Have Users Reacted?

The announcement has sparked backlash online, with many questioning Amazon’s motives. Privacy advocates argue that this is less about AI improvements and more about data collection and monetisation. Some commentators have reacted suspiciously, suggesting that this could be a step towards making Amazon’s ecosystem more reliant on surveillance-driven AI.

On Reddit, frustrated Echo users have been voicing their concerns, with some suggesting it may be time to switch to an alternative. Also, many have pointed to previous scandals, such as when Amazon admitted in 2019 that human reviewers were listening to some Alexa recordings, and in 2023, the company was fined $25 million for storing voice recordings of children.

Privacy-conscious users may now be looking for alternatives, e.g. Apple’s on-device Siri processing or open-source voice assistants like Mycroft AI.

What About Echo Devices in Businesses and Home Offices?

The change doesn’t just impact home users because businesses that use Alexa-powered devices may also need to rethink their approach. For example:

– Conference rooms and smart offices. If companies have Alexa-enabled speakers in workspaces, they now need to consider the implications of sending voice data to Amazon’s servers.

– Home offices. Professionals who use Alexa for reminders, to-do lists, or calendar management may be less comfortable knowing all voice data is processed externally.

– Regulatory compliance. Businesses that handle sensitive information may need to reassess their data security policies, especially in industries like finance, healthcare, and law, where client confidentiality is critical.

Some businesses could choose to disable Alexa in certain settings altogether, particularly where privacy concerns are paramount.

What Are the Wider Implications for Amazon?

Amazon is betting big on Alexa+ and generative AI, hoping to keep pace with competitors like Apple and Google. But this shift also risks alienating privacy-conscious customers.

At the moment, less than 0.03% of Alexa users had the “Do Not Send Voice Recordings” setting enabled, according to Amazon. However, the backlash suggests that even users who didn’t enable this option are now thinking twice about trusting Alexa.

The move could, for example:

– Drive privacy-focused users to competitors like Apple’s Siri, which processes more requests on-device.

– Open the door for legal challenges around how Amazon informs users about data processing.

– Hurt trust in Alexa, at a time when voice assistants are already facing stagnating growth.

While Amazon insists that security remains a priority, the fact remains that this is a fundamental change to how Alexa works, and it’s happening whether users like it or not.

What Does This Mean For Your Business?

While Amazon is framing this change as a necessary evolution to support more advanced AI-driven capabilities, many users clearly see it as a loss of control over their privacy. By removing the ability to keep voice commands on-device, Amazon appears to be fundamentally altering the way Echo devices function, a move that some customers feel undermines the trust they originally placed in the product.

For everyday users, this means having to accept that their voice data will always pass through Amazon’s servers, even if it is not stored long-term. While Amazon has pledged that recordings will be deleted after processing, past controversies involving stored conversations and human reviewers have left some sceptical. Those who put a great deal of emphasis on privacy may now feel compelled to seek alternatives, such as Apple’s Siri, which offers more on-device processing, or open-source options that provide greater transparency.

The implications of this change stretch beyond individual households. For example, UK businesses using Echo devices in offices, meeting rooms, or customer-facing environments may now need to reassess their approach. Many industries, particularly those handling sensitive client data, already have strict privacy policies in place. With all Alexa commands now routed through Amazon’s servers, businesses will need to consider whether continuing to use these devices aligns with their compliance requirements. Some may opt to disable Alexa in certain settings, while others might look for alternative solutions that allow for local voice processing.

From Amazon’s perspective, this change is about keeping Alexa competitive in an AI-driven world. With Google and Apple enhancing their voice assistants, and ChatGPT-like models becoming increasingly integrated into everyday tech, Amazon is pushing Alexa to become more responsive, conversational, and intelligent. However, this ambition looks likely to come at a cost. By prioritising cloud-based AI, Amazon risks alienating a portion of its user base, particularly those who bought Echo devices for their privacy features.

HP has come under scrutiny for implementing a policy that enforced a minimum 15-minute wait time for customers seeking telephone support for consumer PCs and printers.

Frustration

The move, which was (quietly) introduced earlier this year, was designed to push customers towards self-service digital support options, but it has triggered frustration among users and raised concerns about customer service standards.

Internal Memo Reveals Strategic Delay

An internal HP memo, which was leaked to the press, appears to show that the 15-minute call delay was intentional. The document stated that the company wanted to “influence customers to increase their adoption of digital self-solve” and “generate warranty cost efficiencies.”

The changed policy affected customers in the UK, Ireland, France, Germany, and Italy, with HP’s phone system informing callers of the extended wait time before redirecting them to HP’s online support tools, such as its website and virtual assistant. The idea was to encourage users to find answers themselves rather than immediately resorting to human customer service.

Customer Backlash and Internal Dissent

However, unfortunately for HP, its decision didn’t go unnoticed. Customers quickly voiced their frustration, arguing that a forced waiting period was an unreasonable tactic to discourage live support requests. Many took to social media and consumer forums, questioning why a major tech company would deliberately delay assistance.

Internally, the policy was reportedly not well received by HP employees either. It’s been reported that sources within the company suggested that frontline staff had no direct involvement in the decision, leaving them to handle complaints from disgruntled customers without a clear justification.

U-Turn After Widespread Criticism

Facing significant backlash, HP reversed the 15-minute waiting policy within weeks of its implementation. The company issued a statement clarifying that the move was meant to encourage digital solutions and reduce enquiry resolution times, but it acknowledged that customer expectations were not met.

“We are committed to delivering an exceptional customer experience and have listened to our customers’ feedback,” HP said. The company assured users that they would no longer experience artificial delays when calling for support.

Part of a Shift From Traditional Customer Service Models?

For businesses that rely on HP devices, the brief but controversial policy raises important questions about customer support access and service reliability. While HP may have abandoned the delay, the incident highlights an ongoing industry trend of shifting away from traditional customer service models. For example:

– Operational efficiency at risk. Businesses that rely on quick resolutions for IT issues could face significant disruptions if similar policies are reintroduced.

– Rising costs and productivity loss. Longer wait times for technical support translate to delayed troubleshooting, which can impact productivity and profitability.

– Trust and vendor loyalty. Companies may start re-evaluating their relationship with HP, particularly if competitors offer more accessible customer support.

A Growing Industry Trend

HP doesn’t appear to be alone in pushing customers towards digital self-service. For example, many technology companies are investing in AI-powered chatbots and automated support to cut costs. Some firms have even implemented fees for live customer support, reinforcing the idea that human assistance is becoming a premium service.

However, the balance between automation and accessibility remains a concern. A recent Gartner report suggests that consumer protection laws in the EU may soon mandate a “right to human support” to prevent companies from making digital-only assistance the default option.

What Does This Mean For Your Business?

HP’s brief experiment with forced waiting times may have ended, but it appears to have left some lingering concerns over how major tech firms balance cost-cutting with customer care. It seems that businesses should, therefore, remain cautious about future shifts in HP’s support strategy, as the company’s willingness to experiment with such measures suggests a broader trend towards digital-first service models.

While automation can streamline some processes (and cut costs), the need for live support remains critical, particularly in high-stakes business environments where downtime can be costly. Companies should evaluate whether HP’s evolving approach aligns with their operational needs or if alternative vendors offer more reliable support.

At the same time, HP and other industry leaders should probably recognise that restricting access to human assistance could drive customers towards competitors who offer direct service rather than cost-saving efficiencies. The growing tension between digital automation and consumer expectations suggests that future policies will need to strike a careful balance or risk further alienating business customers.

As customer expectations continue to evolve, businesses must be prepared to advocate for accessible and reliable support services, whether from HP or any other technology provider.

A new Nozomi Networks report has found that 94 per cent of Wi-Fi networks lack proper protection against deauthentication attacks, leaving them vulnerable to disruption and infiltration.

The analysis of 500,000+ wireless networks revealed that only 6 per cent had enabled Management Frame Protection (MFP), a key security feature preventing attackers from forcing devices offline using spoofed deauthentication frames. These denial-of-service (DoS) attacks can disrupt connectivity and serve as a gateway for data theft and system intrusions.

Deauthentication attacks exploit weaknesses in Wi-Fi protocols, allowing hackers to disconnect devices, intercept data, and launch man-in-the-middle attacks. Nozomi warns that state-linked hacking groups, such as Volt Typhoon, have used these tactics to target critical infrastructure, healthcare, and industrial networks.

To mitigate risk, businesses should enable 802.11w (MFP), upgrade to WPA3 encryption, and monitor networks for suspicious activity. Regular wireless security audits and strong segmentation are also essential to prevent attackers from exploiting these vulnerabilities.

A new report from the Rhodium Group claims that advanced geothermal energy could power nearly all new data centres by 2030.

A Stable, Renewable Solution

With the exponential rise in artificial intelligence (AI) and cloud computing driving unprecedented demand for electricity, the energy consumption of data centres is a growing concern. However, the report suggests that tapping into the Earth’s heat could provide a stable, renewable solution to this looming energy crisis.

The Data Centre Energy Problem

Data centres are essentially the backbone of the digital economy, hosting everything from cloud storage to AI model training. However, their hunger for power (and water) is becoming a pressing issue. For example, according to the Rhodium Group, electricity demand from data centres in the US has surged from 2 per cent of total consumption in 2020 to around 4.5 per cent in 2024. Projections indicate this could rise to as much as 12 per cent by 2028.

Much of this surge comes from the rapid expansion of AI, with models such as ChatGPT, Google Gemini, and Microsoft Copilot requiring massive computational power. The grid is struggling to keep pace, with utilities and regulators facing growing challenges in ensuring reliable, low-carbon electricity supply.

What Is Geothermal Energy?

Geothermal energy harnesses heat stored beneath the Earth’s surface, converting it into electricity or direct heating. Traditionally, geothermal power plants were limited to areas where hot water or steam naturally rises close to the surface, such as Iceland or parts of the western US.

However, advancements in enhanced geothermal systems (EGS), which use deep drilling and hydraulic fracturing techniques to unlock heat from otherwise inaccessible rock formations, are now changing the game. For example, according to the US Department of Energy, EGS could unlock up to 90 gigawatts (GW) of geothermal capacity in the US alone, providing a vast, untapped source of clean energy.

Why Geothermal Could Meet Data Centre Demand

The Rhodium Group’s report estimates that under current trends, geothermal could provide up to 64 per cent of new data centre electricity demand by 2030. If data centre developers strategically site their facilities in areas with the best geothermal resources, this figure could rise to 100 per cent!

In practical terms, this means geothermal could quadruple its current installed capacity in the US, from 4GW today to approximately 16GW by the end of the decade. Crucially, the cost of geothermal energy is expected to be competitive with existing power sources, ranging from $50 to $75 per megawatt-hour (MWh) (on par with current grid electricity prices for data centres).

Real-World Examples of Geothermal in Action

A number of innovative startups are already proving the feasibility of geothermal-powered data centres. These include:

– Fervo Energy, founded by former oil and gas engineers, has been pioneering horizontal drilling techniques to boost geothermal output. The company secured over $200 million in investment in 2024 and has significantly reduced drilling costs.

– Bedrock Energy is focusing on space-constrained urban environments. Their deep-drilling approach enables office buildings and data centres to tap into geothermal heat with a small footprint.

– Quaise Energy has developed a breakthrough technology that uses high-powered microwaves to vaporise rock, allowing them to drill as deep as 12.4 miles (20km). At these depths, temperatures exceed 1,000°F, providing an almost limitless source of heat.

– Sage Geosystems is taking a different approach, using geothermal wells to store energy. Water is injected under pressure and later released to generate electricity, similar to an underground hydroelectric dam.

The Benefits of Geothermal for Data Centres

Geothermal energy offers a host of advantages for data centre operators, most notably:

– 24/7 reliability. Unlike wind or solar, geothermal provides continuous, baseload power with 90 per cent+ capacity factors.

– A low carbon footprint. Geothermal plants emit little to no greenhouse gases, aligning with tech companies’ aggressive net-zero targets.

– Grid independence. By using behind-the-metre geothermal installations, data centres can bypass lengthy grid connection delays, reducing wait times for power.

– Cost stability. Unlike natural gas, which is subject to price volatility, geothermal energy provides long-term price certainty.

The Challenges of Scaling Geothermal

However, while the potential is pretty clear, there are some significant hurdles to overcome before geothermal can power the next wave of data centres. These include:

– High upfront costs. Deep drilling and well stimulation require significant capital investment. However, costs are falling as technology advances.

– Permitting delays. In the US, for example, securing permits for geothermal projects can take up to 10 years! Streamlining regulatory approvals is crucial to accelerating deployment.

– Geographic constraints. While EGS expands geothermal’s reach, the best sites are still concentrated in the western US, meaning some data centres may have to relocate or use hybrid energy strategies.

– Infrastructure readiness. Drilling rigs, turbines, and skilled labour all need to scale rapidly to meet growing demand. Leveraging expertise from the oil and gas sector could help bridge this gap.

What Does This Mean For Your Organisation?

Geothermal energy could be a real and practical way to address the growing electricity demands of data centres while aligning with sustainability goals. The technology is proven, its reliability is unmatched among renewables, and its potential is vast. However, realising this potential requires significant investment, regulatory reform, and strategic siting of new facilities.

For data centre operators, integrating geothermal could reduce dependence on fossil fuels and offer long-term cost stability. For policymakers, streamlining permitting processes and incentivising geothermal development will be key to unlocking its full potential. Meanwhile, investors and energy companies have a chance to shape a growing market by developing innovative drilling and power generation techniques.

For UK businesses, the need for cleaner, more stable energy sources is just as pressing in this country, where data centre energy demand is rising rapidly. While the UK lacks the geothermal resources of the US, investment in energy innovation, including geothermal heating and advanced drilling techniques, could provide valuable lessons and opportunities. Also, British companies specialising in energy technology, infrastructure, and financing may find growing international demand for their expertise.

The question, therefore, is no longer whether geothermal can support the data centre boom, but how quickly the industry can scale to meet demand. With the right mix of investment, policy support, and technological innovation, the heat beneath our feet could soon be powering the digital world in ways previously unimaginable.

In contrast to the ChatGPT ‘memory’ feature (that we outlined last week), this week’s Tech Talk suggests some ways to use their new ‘Temporary’ feature, in case you want to sandbox your conversations and keep them private.

[Note – To Watch This Video without glitches/interruptions, It may be best to download it first]

When chatting with ChatGPT on WhatsApp, you can now send images and voice messages, allowing the AI to analyse photos and respond to spoken queries. Here’s how to make the most of these new features.

How to Chat with ChatGPT on WhatsApp

– Before using images and voice notes, you’ll need to set up ChatGPT as a contact in WhatsApp.

– Open WhatsApp and tap the New Chat button.

– Select New Contact and enter ChatGPT as the name.

– Under Country, select United States (even though you’re in the UK) and enter the number 1-800-242-8478.

– Save the contact, then open a chat with ChatGPT and start typing your queries.

How to Send Images to ChatGPT

Want AI-powered insights on your photos? You can send images to ChatGPT for analysis.

– Open your chat with ChatGPT in WhatsApp.

– Tap the paperclip or camera icon.

– Choose an existing photo or take a new one.

– Type a message with your query, such as “What’s in this image?” and send it.

– ChatGPT will analyse the image and reply with relevant information.

How to Use Voice Messages with ChatGPT

Prefer to talk instead of type? ChatGPT can now process voice messages and reply in text.

– In your chat with ChatGPT, tap the microphone button.

– Record your question and send the voice message.

– ChatGPT will transcribe your audio and respond with a text reply.

– You can continue the conversation with more voice messages or switch to text.

More Ways to Use ChatGPT on WhatsApp

Need quick help while multitasking? You can also:

– Ask ChatGPT to summarise text or documents by sending images of printed pages.

– Get descriptions of objects, landmarks, or signs by snapping a picture.

– Use voice messages to dictate ideas, notes, or reminders hands-free.

With these new features, interacting with ChatGPT in WhatsApp is more natural and intuitive than ever.