Can AI be trusted with confidential info?
That was the query at stake when well-liked AI transcription service Otter, which has more than 14 million users, despatched an automated transcript of a Zoom assembly to AI researcher Alex Bilzerian final week after he met with a VC agency. The problem is that Bilzerian wasn’t meant to see the total transcript, which additionally transcribed conversations he wasn’t meant to listen to between the VCs after the official name.
Bilzerian wrote that the AI included “hours of their non-public conversations” together with “intimate, confidential particulars about their enterprise” in a post on X that has been seen over 5 million occasions.
A VC agency I had a Zoom assembly with used Otter AI to report the decision, and after the assembly, it routinely emailed me the transcript, together with hours of their non-public conversations afterward, the place they mentioned intimate, confidential particulars about their enterprise.
— Alex Bilzerian (@alexbilz) September 26, 2024
Otter responded to Bilzerian on X and said that the VC firm might have opted to not share the transcript routinely. It’s unclear if the corporate forgot to finish a Zoom name or flip off the AI transcription after the decision.
Bilzerian instructed The Washington Post that he determined not to go through with the deal primarily based on the transcript he acquired.
X customers raised issues about AI’s elevated footprint in conferences and shared associated anecdotes in response to Bilzerian’s submit.
Mark Cecchini, director of wealth options at Quadrant Capital, wrote that AI assistants had been becoming a member of conferences extra usually with out being cleared to take action. Kristen Ruby, CEO of public relations agency Ruby Media Group, pointed out that digital remedy occurs by way of recorded video calls, resulting in a potential information hack.
Associated: These Days, Everything Is ‘Powered By AI.’ Here’s How to Tell Hype From Real Innovation.
One X consumer wrote that his spouse received an AI transcript of a non-public assembly whereas one other claimed they acquired a transcript after an interview with a recorded dialog about why they weren’t match.
AI privateness issues aren’t new. Based on the Middle for Strategic and Worldwide Research, AI has the potential to increase the scope of surveillance by enhancing facial recognition know-how and predictive algorithms on social media.
Nonetheless, AI could have an upside. A Might 2024 Goldman Sachs study estimates that AI might enhance productiveness by 25% and notes that “the early alerts of future productiveness positive aspects look very, very constructive.”
Associated: AI for the Underdog — Here’s How Small Businesses Can Thrive With Artificial Intelligence