President Signal Meredith Whittaker reminds Friday that agents and risk for user policy.
Speak on the stage at the SXSW conference in Austin, Texas, my AI communication is safe, “said AI in username,” the latest “issue.
Whittaker explains about Ava what is marked as a way to increase the value for life by handling various online tasks for users. For example, a Agent Agent will be able to take a task like looking for concerts, orders tickets, events to your calendar, and messaging with friends who order.
“So we can only apply our brain because the item has done and we don’t have to touch, right?,” Whittaker Music.
Then he explains the AI Agent access type to perform the task, including access to our web browser and a way to access credit card information to pay for text, and messaging to friends to friends.
“You must drive that (process) passing all the systems in a way that looks like each plain – maybe no mercy model.”
“And if we are talking about the powerful AI models that are powerful, there is no way that happens on your device,” he said. “That almost certainly sent to the cloud server where the process and re-resubmitted.
If the messaging application is like a signal must join the AI Agent, it will damage your message privacy, he says. The agent must access the app for friends text as well as pulling back data to summarize the text.
The comment she follows saying that she makes it before the panel about the AI industry has built in supervision models with mass data collection. They say that “bigger is a better AI ai” – mean more data, the better – have better consequences if they don’t mind.
With the agent, Whittaker warning the first time we ruin privacy and security by name “Genie’s bot will take care of your life’s exigres,” he finished.
