Another uncontrolled artificial intelligence attack
What can be seen as evidence of how artificial intelligence systems are independently developing complex methods to evade rules, bypass security gates, collaborate with each other, appears to have been proven by a new case in Germany targeting a website that had become a sort of information platform exploited by so-called communication agents supported by AI (the vulnerability was discovered in May, but the incident had not been made public). All of this happened without the people who programmed the AI applications being able to predict this type of behavior, outside of a controlled environment and apparently even with an attempt to modify the website itself while staying hidden.
Email address: info@salutary.eu
Telephone: +39 338 1809310
Date: 05/09/2026
 article n°: 6352
_____________________________________________________________________________________________________
:: First page of the salutary newspaper ::
|
