Last Updated:
AI agents will become a force in the next few years but Signal chief is worried about its impact on our privacy and safety.
AI agents are quickly becoming the toast of the industry but is it all good? (Photo: AI generated)
AI agents have created excitement among people who can’t wait to see its potential. However, president of Signal, the secure messaging app has issued a warning about the potential privacy risks associated with these AI agents. The warning from Meredith Whittaker comes amidst growing depending on AI and machine learning applications. Agentic AI is a type of artificial intelligence that can think, plan, and act independently. It can learn, adapt, and solve problems based on context and objectives.
An advocate of secure communications, Meredith was speaking onstage at the SXSW conference in Austin, Texas. In her speech, she described the use of AI agents as “putting your brain in a jar” and warned that this new computing paradigm, in which AI does almost most of the tasks for users, has a “profound issue” with both privacy and security.
Meredith described how artificial intelligence bots are being promoted as a method to improve human life by taking care of different online duties. AI agents are capable of doing several activities, such as search for concerts, purchase tickets, put the event on a calendar, and notify friends about plans with pre-reservations.
“So we can just put our brain in a jar because the thing is doing that and we don’t have to touch it, right?,” Meredith said, as quoted by TechCrunch. She then went on to describe the kinds of access the AI agent would require to carry out these duties, such as access to a user’s web browser and a means of controlling it, as well as access to their calendar, messaging app, as well as credit card details for payment of tickets and texting friends.
“It would need to be able to drive that (process) across our entire system with something that looks like root permission, accessing every single one of those databases — probably in the clear, because there’s no model to do that encrypted,” Meredith cautioned.
The privacy of a user’s messages would be compromised if a messaging service like Signal integrated with AI agents, Meredith stated. To text friends, the agent must first access the app and then retrieve data to compile the texts.
This is not the first time when the Signal President has been outspoken about her views on AI. Her recent statement came a few days after she discussed how the AI sector was founded on a surveillance strategy that involved the collecting of large amounts of data. She expressed her views on a panel while cautioning users about how the AI has been misusing and controlling user data.
