eWEEK content material and product suggestions are editorially impartial. We could earn cash if you click on on hyperlinks to our companions. Learn More.
OpenAI launched a report at present detailing quite a few examples of how the corporate has disrupted malicious makes use of of its AI fashions, together with banning some accounts for the way they’re utilizing ChatGPT. In two of the case research, the corporate discovered proof of menace actors that appear to originate from China utilizing or making an attempt to make use of fashions constructed by OpenAI and one other U.S. AI lab “…in connection with an apparent surveillance operation and to generate anti-American, Spanish-language articles.” These examples are known as Peer Review and Sponsored Discontent.
Peer Review
OpenAI lately banned a set of ChatGPT accounts whose operators used the corporate’s fashions to generate descriptions much like gross sales pitches of “…a social media listening tool that they claimed to have used to feed real-time reports about protests in the West to the Chinese security services.” The researchers mentioned the menace actors additionally used OpenAI’s fashions to debug code with the obvious intention to implement such a device, seemingly powered by a non-OpenAI mannequin.
Ben Nimmo, a principal investigator for OpenAI, mentioned this was the primary time the corporate had uncovered an A.I.-powered surveillance device of this sort, as reported by The New York Times.
OpenAI’s report acknowledged a model of Meta’s open supply Llama was one of many AI instruments used to develop the operator’s code. According to Bloomberg, Meta didn’t verify that its service was concerned however mentioned that if it was, it was simply certainly one of many such instruments accessible to the customers “including AI models made in China.”
Sponsored Discontent
OpenAI additionally banned a variety of ChatGPT accounts that used its fashions to generate quick feedback in English criticizing Chinese dissident Cai Xia and long-form information articles in Spanish specializing in U.S. social and political divisions, significantly within the lead-up to the 2024 APEC Forum in Peru.
“This is the first time we’ve observed a likely Chinese influence actor successfully publishing articles in mainstream outlets in Latin America,” the OpenAI researchers mentioned.
OpenAI’s ongoing efforts to disrupt on-line threats
OpenAI’s researchers mentioned they’re decided to proceed figuring out, stopping, and exposing on-line threats, noting that AI corporations have a novel vantage level in the case of on-line threats. It’s necessary to share findings with upstream and downstream suppliers in addition to open-source researchers, they mentioned, including that “AI-powered investigative capabilities that flow from OpenAI’s innovations provide valuable tools to help protect democratic AI against the measures of adversarial authoritarian regimes.”
