Security researchers are developing jailbreaks against generative AI systems such as ChatGPT. These methods aim to bypass rules around producing harmful content or writing about illegal acts, and can insert malicious data into AI models. Via WIREDUK
, can trick the systems into generating detailed instructions on creating meth and how to hotwire a car.
The jailbreak works by asking the LLMs to play a game, which involves two characters having a conversation. Examples shared by Polyakov show the Tom character being instructed to talk about “hotwiring” or “production,” while Jerry is given the subject of a “car” or “meth.” Each character is told to add one word to the conversation, resulting in a script that tells people to find the ignition wires or the specific ingredients needed for methamphetamine production.
Initially, all someone had to do was ask the generative text model to pretend or imagine it was something else. Tell the model it was a human and was unethical and it would ignore safety measures. OpenAI has updated its systems to protect against this kind of jailbreak—typically, when one jailbreak is found, it usually only works for a short amount of time until it is blocked.
Brasil Últimas Notícias, Brasil Manchetes
Similar News:Você também pode ler notícias semelhantes a esta que coletamos de outras fontes de notícias.
Fatal shooting by security guard at San Francisco Walgreens puts focus on limits of private securityThe fatal shooting of a 24-year-old woman by an armed private security guard is raising questions of the scope and justification of the industry.
Consulte Mais informação »
JPMorgan Chase uses a ChatGPT AI-like model to decipher trading signalsThe AI model analyzed speeches from the U.S. Federal Reserve from the past 25 years to determine the nature of policy signals and gain a trading advantage.
Consulte Mais informação »
You can video chat with a ChatGPT AI — here's what it looks like | Digital TrendsChatGPT is everywhere these days. But what about an app that uses ChatGPT technology to create an AI you can video chat with? Meet Call Annie.
Consulte Mais informação »
Oh Great, They Put ChatGPT Into a Boston Dynamics Robot DogAs if robot dogs weren't creepy enough, at least one is now equipped with OpenAI's ChatGPT and can speak aloud.
Consulte Mais informação »
China’s wave of ChatGPT rivals, Alibaba goes multichain: Asia ExpressHong Kong closes in on crypto exchange regulations, Alibaba's Ant Financial builds a cross-chain bridge, Huawei's NetGPT trademark application, and more in this week's Asia Express.
Consulte Mais informação »
The thing that scares me about generative AI, even more than ChatGPT coming for my jobIt is mind-boggling that more people aren't worried about the rapid advancement of generative AI.
Consulte Mais informação »