The UK is sounding the alarm.
British officials are warning companies against integrating chatbots powered by artificial intelligence, as research has shown that they can be tricked into performing malicious actions. In two blog posts published on Wednesday, the UK's National Cybersecurity Center (NCSC) pointed out that experts have not yet evaluated all the potential security problems associated with algorithms capable of creating human – like interactions-the so-called large language models (LLMs).
Artificial intelligence-based tools are beginning to be used early as chatbots, which some believe will replace not only Internet search, but also customer service and sales calls. NCSC warned about the risks, especially if such models are integrated into other business processes of the company. Researchers have found that chatbots can be tricked by sending malicious commands to log in or bypassing their built-in security mechanisms. For example, a chatbot used by a bank may be deceived to perform an unauthorized transaction.
NCSC in one of its posts states: "Organizations developing LLM-based services should exercise caution, as if they were using a product or code library in beta testing. You need to approach LLM with the same caution."
Authorities around the world are grappling with the growing number of LLM programs, such as OpenAI's ChatGPT, which companies include in a wide range of services, including sales and customer service. The impact of artificial intelligence on security is also still in the spotlight, with US and Canadian authorities claiming to have seen hackers using the technology.
Oseloka Obiora, CTO of cybersecurity firm RiverSafe, said the race to integrate artificial intelligence into business practices will have "catastrophic consequences" if business leaders do not introduce the necessary checks.
"Before following the latest trends in AI, top managers should re-evaluate the situation, weigh the pros and cons, and implement the necessary cyber defense measures," he said.
British officials are warning companies against integrating chatbots powered by artificial intelligence, as research has shown that they can be tricked into performing malicious actions. In two blog posts published on Wednesday, the UK's National Cybersecurity Center (NCSC) pointed out that experts have not yet evaluated all the potential security problems associated with algorithms capable of creating human – like interactions-the so-called large language models (LLMs).
Artificial intelligence-based tools are beginning to be used early as chatbots, which some believe will replace not only Internet search, but also customer service and sales calls. NCSC warned about the risks, especially if such models are integrated into other business processes of the company. Researchers have found that chatbots can be tricked by sending malicious commands to log in or bypassing their built-in security mechanisms. For example, a chatbot used by a bank may be deceived to perform an unauthorized transaction.
NCSC in one of its posts states: "Organizations developing LLM-based services should exercise caution, as if they were using a product or code library in beta testing. You need to approach LLM with the same caution."
Authorities around the world are grappling with the growing number of LLM programs, such as OpenAI's ChatGPT, which companies include in a wide range of services, including sales and customer service. The impact of artificial intelligence on security is also still in the spotlight, with US and Canadian authorities claiming to have seen hackers using the technology.
Oseloka Obiora, CTO of cybersecurity firm RiverSafe, said the race to integrate artificial intelligence into business practices will have "catastrophic consequences" if business leaders do not introduce the necessary checks.
"Before following the latest trends in AI, top managers should re-evaluate the situation, weigh the pros and cons, and implement the necessary cyber defense measures," he said.
