4 min read
Over five years ago, The Guardian revealed allegations that Apple used private, unintentionally recorded communications to train Siri without users’ consent. Now, Apple has agreed to a $95 million settlement to resolve claims over “unintended Siri activation,” clearing it of further damage claims (Source: Ars Technica).
The issue surfaced in June 2019, following similar admissions from Amazon and Google about unintentional recordings. If approved by the US District Court, the settlement will offer compensation to those who owned Siri-enabled devices between September 17, 2014, and December 31, 2024. As part of the deal, Apple must confirm the deletion of Siri recordings before October 2019 and update its guidelines on voice data use, including the opt-in process for anonymized submissions to “Improve Siri.”
Proof of Voice Assistant Espionage?
It was agreed that Apple and related companies would not be responsible for the claimed breach after the payment. In essence, from 2014 until 2024, the corporation and all individuals involved would be absolved of responsibility and required to deny any wrongdoing in relation to what could have potentially cost Apple more than $1.5 billion under the Wiretap Act.
Apple’s apparent release from responsibility contradicts the 2019 public outrage. Users, industry experts, and journalists screamed foul over the alleged exploitation of recordings that were captured unintentionally and could not have been consented to. However, as early as 2013, Apple freely disclosed to Wired their two-year Siri data retention policy, which included anonymity after six months. However, Apple never actually admitted in that policy that it had received accidental recordings.
The ‘Magic’ Behind Voice Recognition Revealed
In 2018, Amazon faced backlash after a private conversation was inadvertently recorded by Alexa and sent to a user’s contact. The company attributed the incident to a misunderstood sequence of commands, explaining that a background word resembling “Alexa” had triggered the assistant. Alexa then misinterpreted additional background chatter as a “send message” command. While this chain of events was improbable, it was deemed accidental rather than an act of intentional eavesdropping on the couple’s conversation about hardwood flooring.
A year later, in 2019, Apple’s whistleblower revelations surfaced, overshadowing another recent controversy. Just weeks earlier, over 1,000 Google Assistant recordings were leaked to the Belgian news outlet VRT NWS. In response, Google temporarily halted the use of voice recordings and faced a three-month suspension imposed by German regulators. An internal investigation eventually placed the blame on third-party contractors responsible for the leaks.
After this, both Apple and Google promised to increase security, eliminate default data retention, and cut ties with third-party contractors. Even before Apple was accused, whistleblowers had exposed Amazon in April for granting third-party employees access to customer recordings. A goldfish’s memory was better than the general population’s.
Are your private conversations being used for ads?
When one considers how a business might initially train voice recognition, the initial concerns about covert recording promptly fade. Speaking about how training voice recognition software on user recordings is “business as usual” in 2019, Forbes contributor Kevin Murnane made the remarks. The “evil tech” rhetoric that surrounded the various recording discoveries was “ridiculous,” Murnane continued, as he outlined Google’s internal troubles.
Although the $95 million payout is negligible, customers have already jumped on the notion that Apple is thwarting a huge plot. “Two plaintiffs said their mentions of Air Jordan sneakers and Olive Garden restaurants triggered ads for those products,” Reuters says, a completely unproven assertion that shows how paranoid thinking causes us to ignore facts and significant problems.
What’s the bigger picture beyond the spying claims?
Headlines like “Alexa has been eavesdropping on you this whole time” in the Washington Post aren’t telling you the truth about how companies and marketers really track users. Companies actually track us through cookies, browser fingerprinting, and “ghost profiles.” These are much more useful and harder to stop than secretly turning on a microphone, sending audio to a server, parsing and sorting information, and then showing it to users through highly personalized ads.
According to the settlement papers, Apple offered $95 million because the business did not see any benefit in continuing the litigation in the uncertain and rapidly evolving field of digital privacy. No voice assistant consistently correctly identifies wake words, yet every speech recognition developer trains with user inputs. Tech companies must keep tightening their security procedures since accidental recordings will always occur. However, more conspiracies are unnecessary.
