In a blog post, the company made sure to say that audio, video, and chats were not used without permission for AI.
Users of the video-calling app took action when the company’s terms of service changed in March in a way that made them worry that AI could be trained.
The company said it changed things to be more open.
In June, Zoom added new AI-powered features. One of these lets users summarize talks so they don’t have to record the whole thing. People could try out the tools for free.
But some experts said that the way the terms of service were written at first could have given Zoom access to more user information than was necessary, including information from customer talks.
Before the terms of service were changed, Robert Bateman, a data security expert, told the BBC, “The terms seemed to give the service provider a lot of freedom to use data made by its users for many different reasons.”
Even though he didn’t know for sure what risks could happen, he said, “alarm bells should go off when you see broad contractual provisions like these.”
Late on Monday, Zoom changed its terms to say “Zoom will not use audio, video, or chat customer content to train our AI models without your permission.”
Human-like behaviour
AI applications are computer programs or tools that can do certain intelligent jobs that humans would usually do. They are taught to “learn” and act like humans by using huge amounts of data and complicated formulas.
But the mass collection of online information to train the models that power AI apps has raised concerns and led to lawsuits over the possibility that their datasets contain private, sensitive, or copywritten information.
Zoom, like many other tech companies, has put more focus on AI goods this year in response to the growing buzz around the technology.
But the Open Rights Group, which works to protect digital privacy, says that the changes are “more alarming” because Zoom is offering the new features as a free trial and encouraging users to “opt in.”
“While Zoom’s privacy policy says that customers will be asked for permission to use their data to train AI models, it is not clear that this is the case,” policy manager for data protection Abby Burke told the BBC before the latest change to Zoom’s terms.
On Monday, a Zoom representative said again that it was up to the customer to decide whether or not to use generative AI features and, separately, whether or not to share customer material with Zoom to “improve the product.”
In Monday’s blog post, screenshots showed examples of warning messages for people who use AI tools to join meetings, giving them the choice of agreeing to the training use or leaving the meeting.
Zoom’s chief product officer, Smita Hashim, said that account owners and administrators could choose to turn on the features, which were still in a trial phase, and that people who did so would “be presented with a transparent consent process for training our AI models using your customer content.”
