UK intelligence agency warns users about dangers of AI chatbots like ChatGPT

British Intelligence Agency has warned about the potential dangers posed by AI Chatbots like ChatGPT and other large language model-based chatbots (LLM).

  • Facebook Share Icon
  • Twitter Share Icon
  • WhatsApp Share Icon
 
Follow : Google News Icon
AI Chatbots
Image: Unsplash (Representative) | Image: self

British Intelligence Agency has warned about the potential dangers posed by AI Chatbots like ChatGPT and other large language model-based chatbots (LLM). In a blog post, the National Cyber Security Center warned the users from entering personal or sensitive information into such software and asked the users to proceed with caution. The cyber security experts urged users to be concerned about privacy leaks to cybercriminals.

As per the blog post, the LLMs are developed and trained using huge datasets and once the information is registered, it can be used to answer similar queries. This kind of training helps the developers to further improve the AI tools. However, this information can turn out to be a major security hazard if they get into the hands of cybercriminals. The criminals can then use the data in a malicious manner. On ChatGPT, which is one of the fastest-growing software in the world, millions of users enter different sets of queries on the AI chatbot. 

Stay cautious about public LLMs

The two major pieces of advice the UK intelligence agency gave in this regard are not to include sensitive information in queries to a public LLM and not to submit queries that would lead to issues if made public. The British centre also revealed that these overexposed queries will help scammers create more convincing and phishing emails and will assist attackers in writing more advanced malware than they have previously attempted. "There have been some incredible demonstrations of how LLMs can help write malware. The concern is that an LLM might help someone with malicious intent (but insufficient skills) to create tools they would not otherwise be able to deploy," the UK intelligence agency asserted in the blog post. However, the centre also had nice things to say about the new innovation. “LLMs are undoubtedly impressive for their ability to generate a huge range of convincing content in multiple human and computer languages.” But made it clear that greater access to information the hands which are unknown can be a matter of concern.

Published By:
 Bhagyasree Sengupta
Published On: