MEREDITH WHITTAKER signal chairman warned that Agent AI could be at risk to user privacy.
While speaking on the stage at the SXSW conference held at Texas Austin, a safe communication advocate mentioned the use of AI agents as “putting AI” agents in the brain, and this new computing paradigm (AI is a new paradigm of doing work on behalf of users) is a “profound problem” for privacy and security. I warned that I had.
Whittaker explained how the AI agent is being sold in a way that adds value to life by handling various online tasks for users. For example, the AI Agent can do a concert, take a ticket booking, the calendar, and send the schedules for the friends to send the reserved friends by message.
“So we can put our brain into a jar because we are doing it.
Then the AI agent explained how the AI agent explained the type of access that needs to do this, including how the AI agent can pursue access to the web browser, and explained how to send texts, calendars and messaging apps to friends by accessing credit card information.
“You must be able to access all systems that can be accessible to that database from all systems that appear to be the same as root privileges. Perhaps there is no encrypted model, so you can see clearly.
“And we are powerful enough…” This is almost certainly transmitted and processed to the cloud server and is transmitted again. Therefore, there is a deep problem in security and personal information protection, which is harassing this over -advertising for agents, which combines all of these separate services to break the blood brain barrier between the application layer and the OS layer. I’m threatening. ”Whittaker concluded.
She said that the message apps such as the signal will damage the personal information of the message if they are integrated with the AI Agent, she said. Agent accesses the app to send a text to a friend and take out the data to summarize the text.
Her opinion was in the panel about a method based on a surveillance model that the AI industry was collected. “The bigger AI paradigm is better,” she said. The more data you have, the better the potential result you think it is worse.
Along with Agent AI, Whittaker warned that it would further weaken personal information protection and security under the name of “Magic Genie Bot to take care of the emergency of life.”
