FBI has an ‘AI’ warning for you – Times of India

The agency expressed concerns during a call with journalists. It revealed that AI chatbots have been used for various illicit activities, including scammers and fraudsters perfecting their techniques and terrorists consulting the tools on how to launch more destructive chemical attacks.
During a conference call with reporters, an anonymous official from the bureau revealed that there are two main risks associated with the use of AI.

The first risk is “model misalignment,” which is when AI software is developed or deployed in a way that leads to undesirable outcomes. The second risk is the direct “misuse of AI” to support other operations.
An official stated that they anticipate an increase in the use of AI models among people as the adoption and democratisation of such technology progresses. Unfortunately, there are also bad actors who can use AI to enhance their criminal activities.
Cybercriminals are using AI to create new malware attacks and delivery methods, such as AI-generated phishing websites and polymorphic malware that can bypass antivirus software. The FBI warns of scammers using AI to create sexually explicit deep fakes for extortion. Criminals also enhance traditional scams, like using AI voice-cloning technology in scam phone calls.
The FBI did not disclose what AI models criminals are using. However, an official mentioned that hackers are more gravitated towards free, customizable open-source models, as well as private AI programs developed by hackers, which are available in the cybercriminal forum
The officials also mentioned that foreign actors are increasingly targeting and collecting information from US companies, universities, and government research facilities for AI advancements. This includes algorithms, data expertise, computing infrastructure, and even talented individuals.