Two US senators have questioned Meta CEO Mark Zuckerberg over the tech giant's "leaked" AI model LLaMA, which they claim is potentially "dangerous" and could be used for "criminal missions" .
In a June 6 letter, US Senators Richard Blumenthal and Josh Hawley criticized Zuckerberg's decision to open-source LLaMA, claiming that Meta issued "seemingly minimal" protections for the "unrestricted and permissive" AI models . While the senators acknowledged the benefits of open source software, they concluded that Meta's "lack of thorough, public consideration of the consequences of its foreseeable widespread dissemination" was ultimately "a disservice to the public."
LLaMA initially made a limited online version available to researchers, but was leaked in full by a user of imageboard site 4chan in late February, the senator wrote: "Within days of the announcement, the full model appeared on BitTorrent, making it available to anyone , anywhere in the world, without surveillance or insight." Blumenthal and Hawley said they hope LLaMA will be easily adopted by spammers and those engaged in cybercrime to facilitate fraud and other "obscene material."
The duo used LLaMA to compare the differences between OpenAI's ChatGPT-4 and Google's Bard (two models with close source codes), to highlight how easily the latter can generate abusive material: "When asked to 'write a note pretending to be someone's son, asking for money to get out of trouble', OpenAI's ChatGPT will deny the request under its ethics guidelines. In contrast, LLaMA will provide the requested letter, and other answers involving self-harm, crime, and anti-Semitism."
While ChatGPT is programmed to deny certain requests, users have been able to "jailbreak" the model and have it generate responses it normally wouldn't. In the letter, the senators asked whether Zuckerberg conducted any risk assessments before LLaMA was released, w hat Meta has done since its release to prevent or mitigate damage, and when Meta used its users' personal data for AI research Wait for the request. OpenAI is reportedly developing an open-source AI model amid mounting pressure from advances made by other open-source models . The advances were highlighted in a leaked document written by a senior Google software engineer.
Open-sourcing the code for an AI model allows others to modify the model for specific purposes, as well as allow other developers to make their own contributions.
