Meta has found a new source of training data for its AI models: its own employees. The company plans to use data collected from employees’ mouse movements and keystrokes to build more capable and efficient artificial intelligence.
The story, first reported by Reuters, shows how long tech companies will look for new sources of training data – a lifeblood for AI models that help programs learn how to perform tasks more effectively and respond to user queries.
When TechCrunch asked for comment, a Meta spokesperson said: “If we are building agents that help people use their computers to complete everyday tasks, our model will allow people to actually Use mouse movements, button clicks, and drop-down menu navigation. To help with this, we’re releasing internal tools that capture these kinds of inputs from specific applications to help train models. Safeguards are in place to protect sensitive content and the data will not be used for any other purpose.”
This trend exposes a troubling privacy aspect of the AI industry. There were reports last week that older startups are being collected and converted into AI training data for corporate communications (such as Slack archives and Jira tickets).
