An accident with Otter’s AI raises critical questions regarding the management and protection of sensitive information. Alex Bilzerian, a researcher and engineer, recently faced a situation that perfectly illustrates these problems.
Otter AI: automatic transcription in question
Alex Bilzerian used Otter AI, a popular automatic transcription tool, to record a Zoom meeting with venture capitalists. After leaving the meeting, he received an email containing not only the meeting minutes, but also the meeting times private conversations between investors. These discussions included confidential details about their company, leading Bilzerian to abandon any deal with them, fearing a failure to protect sensitive information.
This case is not isolated. In fact, other users have also reported similar incidents where sensitive information was accidentally shared due to incorrect settings or an imperfect understanding of the tool’s capabilities. Tools like Otter AI can unintentionally disclose company secrets or sensitive discussions, thereby increasing the risk of legal action or serious breaches of trust.
The speed of technological development: a challenge for control and understanding
Bilzerian’s case highlights a growing problem with artificial intelligence technologies: They evolve faster than our ability to control or fully understand them. Not everyone who uses these tools is tech savvy. It is therefore critical that companies understand and master the parameters and implications of these tools before adopting them on a large scale.
Common practices of virtual assistants like Otter AI, including the ability to automatically send recordings and transcripts to participants even after the meeting ends, pose significant data privacy and security challenges. In response to these concerns, Otter AI insists that users have a full control over sharing settings conversations and can change, update or stop these shares at any time.
Promising but imperfect tools for improving productivity
Despite these issues, the features offered by AI-based virtual assistants attract many companies looking to improve their productivity. Salesforce recently launched Agentforce, an AI offering for creating virtual agents to help with sales and customer service. For its part, Slack integrates artificial intelligence functions to summarize conversations, search for topics and create daily summaries.
However, the adoption of these technologies must be accompanied by good understanding of their parameters and their implications for avoiding unwanted automatic behaviors. For example, the “auto-transcription” feature may continue to work even after participants leave a meeting, posing a potential privacy breach.
Despite the valuable help that AI can provide in the workplace, it is human skills that remain essential to ensure the correct and ethical use of these tools. Although technology evolves rapidly, humans must be able to effectively supervise and regulate its use to avoid unfortunate incidents.