Come March 28, those who opted to have their voice commands for Amazon’s AI assistant Alexa processed locally on their Echo devices will lose that option, with all spoken requests pushed to the cloud for analysis.

Amazon hasn’t formally announced the change, and the help page for the feature still makes no mention of the March 28 deprecation. But the internet souk confirmed to The Register that emails to users about the update, which caused a stir on social media over the weekend, are indeed legit.

“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,” a copy of the email sent to Echo users relayed to El Reg read.

“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.”

All your phrase are belong to us

So there it is, apparently: Alexa’s latest generative AI tricks are too demanding for the hardware on the handful of Echo devices that support local processing — the 4th-gen Echo Dot, Echo Show 10, and Show 15. Less powerful Echo gadgets don’t have any option for processing locally. Therefore, all spoken Alexa requests are going up into Amazon systems for remote processing.

Privacy-conscious users who enabled the “Do Not Send Voice Recordings” setting won’t get a say; it’s being disabled automatically. Not that many people bothered enabling the local option in the first place, Amazon claims.

“If you do not take action, your Alexa Settings will automatically be updated to ‘Don’t save recordings,'” Amazon told affected users.

“Starting on March 28, your voice recordings will be sent to and processed in the cloud, and they will be deleted after Alexa processes your requests,” the email continued. “If your voice recordings setting is updated to ‘Don’t save recordings,’ voice ID will not work and you will not be able to create a voice ID for individual users to access more personalized features.”

In other words, unless you let Amazon store your recordings, you’re stuck with a feature-limited Alexa. And all voice commands are going up regardless.

Private, for a given definition of privacy

While Echo owners may not be happy about about losing the option for on-device audio processing, the soon-to-be-scrapped feature wasn’t exactly airtight with its privacy protection to begin with.

Another Amazon help page that gives more detail on the Do Not Send Voice Recordings option notes that even when audio recordings stay local, a text transcript of each request still gets shipped off to Amazon’s cloud for processing anyway. Those transcripts are stored right alongside voice recordings and don’t auto-delete — you have to manually purge them via your Voice History, assuming you knew they existed in the first place.

Amazon customers shouldn’t be surprised, though: The tech titan’s approach to privacy has long raised eyebrows, particularly when it comes to Alexa and the other gadgets it plants in homes to gain an audio and visual foothold.

Studies have claimed that Amazon uses Alexa voice interaction data to help target ads — both on Echo devices and across the web. Third-party apps available for Alexa-enabled devices don’t offer much comfort either, at times lacking clear privacy policies or adequate safeguards on how user data is handled.

Then there’s last year’s drama in America surrounding Ring cameras, when the FTC claimed the super-corp’s lax security controls allowed Amazon employees and contractors to access customers’ private video feeds. The agency also alleged that Amazon unlawfully retained Alexa voice recordings of children indefinitely, violating child privacy laws. 

Amazon, naturally, denies that eliminating the on-device processing feature will impede user privacy. 

“The Alexa experience is designed to protect our customers’ privacy and keep their data secure, and that’s not changing,” an Amazon spokesperson told us. “We’re focusing on the privacy tools and controls that our customers use most and work well with generative AI experiences that rely on the processing power of Amazon’s secure cloud.”

It’s those generative AI updates, revealed in late February alongside the launch of Alexa+, that appear to be driving the change. The three devices mentioned by Amazon as supporting local voice processing (Dot 4, Show 10, Show 15) are all in the lineup of devices supported by Alexa+, which Amazon is no doubt keen to push users to adopt. 

Unlike the classic Alexa, which Amazon said will continue to be available, generative AI through Alexa+ – and the fresh stream of user data required to fuel them – will only be available to Amazon Prime subscribers, or anyone willing to shell out $19.99 per month without Prime. Whether you’re sticking with Alexa or using Alexa+, commands are processed remotely.

Amazon told us customers will still have plenty of privacy options available, “including the option to not save their voice recordings at all.” That feature, as we noted above, means losing out on many essential Alexa features, like the voice assistant being able to recognize an individual speaker and respond based on their preferences, which for many multi-user households is an essential feature. 

“We’ll continue learning from customer feedback, and building privacy features on their behalf,” Amazon said. ®

Speaking of AI… Under the direction of President Donald Trump and his US government standards body NIST, scientists linked to the federally funded Artificial Intelligence Safety Institute have been told to talk less about “AI safety,” “responsible AI,” and “AI fairness,” and focus more on “reducing ideological bias, to enable human flourishing and economic competitiveness,” in machine-learning research, WiReD magazine reports.