Search AI Products and News

IBM researchers have discovered a relatively simple method for hijacking voice calls using generative AI tools. This finding raises concerns for organizations that rely on phone verification for identity, such as financial institutions. It poses a challenge for chatbots, as low-cost AI tools enable scammers to easily impersonate others’ voices, stealing funds and other sensitive information. Audio hijacking is a new threat, where malicious actors can manipulate large language models using voice cloning during ongoing conversations.

IBM's research found that phishing emails generated by ChatGPT are deceptive. Although the click-through rate is slightly lower than human-generated emails, ChatGPT produces them much faster. IBM hackers indicate that artificial intelligence is continuously advancing and is expected to surpass human capabilities.

Research shows that executives anticipate 40% of employees will need to reskill to adapt to the rise of artificial intelligence and automation. Junior employees are expected to face the most significant changes, with 77% of executives believing they will be influenced by generative AI. Additionally, 87% of executives think employees are more likely to be augmented by AI rather than replaced.

{1: IBM Research found that large language models such as GPT-4 can be easily deceived to generate malicious code or provide false security recommendations. 2: Researchers discovered that a basic understanding of English and some background knowledge about the model's training data is sufficient to easily deceive AI chatbots. 3: Different AI models vary in their susceptibility to deception, with GPT-3.5 and GPT-4 being more easily tricked.}