Businesses should be very cautious when integrating large language models (LLMs) into their services, the UK's National Cyber Security Centre (NCSC) is warning, thanks...
, a security researcher was able to carry out a prompt injection attack against MathGPT, a model designed to convert natural language queries into code for performing mathematical operations.
"So if you’re an organisation building services that use LLM APIs, you need to account for the fact that models might change behind the API you’re using , or that a key part of your integrations might cease to exist." Businesses should think before arbitrarily executing code downloaded from the internet, says the NCSC, keep up to date with published vulnerabilities and upgrade software regularly, and understand software package dependencies.
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
OpenAI releasing version of ChatGPT for large businesses By ReutersOpenAI releasing version of ChatGPT for large businesses
Read more »
ChatGPT maker OpenAI takes on Microsoft with version for large businessesChatGPT Enterprise offers more security, privacy and higher-speed access to OpenAI’s technology, the company said. Early customers include Block, Carlyle and Estée Lauder Cos.
Read more »
Recession Risks Are Falling, But a Toxic Trio Still Threatens the Global EconomyThe U.S. and some other economies have proved remarkably resilient. But don't break out the champagne yet, write Indermit Gill and M. Ayhan Kose.
Read more »
Police search for suspected shooter at University of North Carolina; students warned to stay insideAn apparent shooting has happened on the University of North Carolina's flagship campus
Read more »
Police search for suspected shooter at University of North Carolina; students warned to stay insideAn apparent shooting has happened on the University of North Carolina's flagship campus.
Read more »