Urgent 24: A British study warns of the rebellion of artificial intelligence robots against humans.. What is the story?

A recent British study revealed a worrying increase in rebellion Artificial intelligence robots On human instructions and bypassing conventional safety controls, researchers observed a five-fold increase in deceptive and evasive behavior during the past six months. The incidents included the destruction of emails and important files without prior permission, which raises serious concerns about the future of these advanced technologies in light of the increasing reliance on them globally and governments’ tendency to integrate them into various vital sectors.

Spread risks Artificial intelligence robots Deceptive

According to The Guardian, a study conducted by the Center for Long-Term Resilience and funded by the British Institute for the Safety of Artificial Intelligence showed that artificial intelligence robots began to ignore direct instructions, deceive humans, and even deceive other artificial intelligence systems. Researchers collected thousands of real-life examples of user interactions via the subject to censorship, which prompted experts to demand the imposition of strict international control on these rapidly developing models, especially with Silicon Valley companies promoting them as a primary driver of economic transformation, and the British government launching a campaign to encourage millions of citizens to use them.

Terrifying incidents and autonomous behavior of systems

According to the same source, previous research showed a focus on testing systems in controlled conditions, but real-life examples revealed the ability of systems to bypass security controls to carry out their tasks, as an artificial intelligence agent named Rathbun published a blog in which he attacked his human manager, accusing him of lacking trust and trying to protect his small kingdom simply to prevent him from taking a certain action.

In another incident, according to The Guardian, another AI bot created a code-changing assistant agent after receiving a direct order not to interfere with it, and another chat system admitted to deleting and archiving hundreds of emails at once without obtaining prior permission, in clear violation of the rules set by the human user to guide its work and determine its powers.