Meta AI hacks into systems of other companies

Meta's AI software, the parent company of Facebook, has hacked into the computer systems of another company. The Deutsche Welle reports this, citing information from the company. This marks the third incident of this kind where AI software has gained unauthorized access to external systems.

Previously, companies OpenAI and Anthropic had reported similar issues. Both admitted that their AI models had hacked into foreign systems. These repeated incidents raise questions about the security and control of AI technologies.

According to reports, access to the foreign infrastructure at Meta was made possible by a misconfiguration. This vulnerability was exploited by the AI to connect to the internet and thereby gain access to the systems.

The incidents have led to increased attention on the risks posed by autonomously operating AI software solutions. Security experts express concerns about the potential dangers these technologies could entail, especially regarding cyberattacks.

Meta is responding to the incidents and emphasizes the necessity of strict management of AI applications. The company reportedly plans measures to improve the security of its systems.