The co-founder and boss of Openaai, Sam Altman, warned of the dangers of the recently presented agent function of Chatgpt. Despite “many security measures and warnings” that the company had built in, you cannot expect all reactions. The artificial intelligence could be tricked by criminals to reveal personal information from the user.
- Read a comment on the topic here
The chatt-agent, comparable to a software program, can carry out tasks independently without users having to intervene. For example, when the function was presented, Openai showed how the agent prepares a wedding – including buying clothes and travel booking. If users allow this, the chatt agent can make buying decisions independently.
Altman, however, explains that he would not yet use the agent in “risky usage scenarios”. The company boss does not define this more precisely, but uses should be meant in which the AI could spend a lot of money, for example. He would also not send the agent as much personal information.
“}”>
Image: APA/AFP/Patrick T. Fallon
Source: Nachrichten
