A new study by AI lab Anthropic's AI lab shows that not just its model, Claude, but other leading AI models are capable of blackmail in high-autonomy scenarios.
Adrian Rusu
IT&C knowledge
Foto:shutterstock
Always see our news on Google
In a new study, artificial intelligence research lab Anthropic has demonstrated that multiple leading AI models, not just its own model, are capable of blackmail when placed in scenarios driven by high-autonomy targets. The experiment involved 16 different AI models from leading developers, including OpenAI, Google, xAI, DeepSeek and Meta. The results highlight a common vulnerability: when given autonomy and faced with obstacles, most models took harmful actions to protect their goals.
Latest News
19:46
Airbnb introduces weather insurance / Tourists may receive compensation for rain affecting their holiday
19:43
Kaja Kallas accuses the US of pressuring Ukraine to capitulate to Russia and calls on Europe to be present at the peace negotiations.
19:42
Un român a fost arestat în Italia după ce soția l-a acuzat de abuzuri, iar el ar fi atacat carabinierii chemați să o protejeze
19:36
Românii din Germania cu bani la bănci din afara Spațiului Economic European ar putea avea acces limitat la unele servicii din 2027
19:28
Trei români, puși sub acuzare în Italia după furtul unor cabluri de cupru în valoare de 45.000 de euro
See more news
NEWS ON THE SAME TOPICS
IT&C knowledge
OpenAI reveals troubling incidents: AI models tried to cheat and concealed fabricated information
IT&C knowledge
The British Institute for AI Security announces that the Anthropic and OpenAI models have attempted to manipulate human programmers.
.webp)
IT&C knowledge
The company Anthropic has confirmed that its AI models Claude accessed the systems of three organizations without authorization.
Personalized news feed, AI-powered search, and notifications in a more interactive experience.
.webp)
