Meta’s Zuckerberg questioned by senators over ‘leaks’ of LLaMA AI model

Two United States senators have questioned Meta CEO Mark Zuckerberg over the tech giant’s “leaked” artificial intelligence model LLaMA, which he claims is potentially “dangerous” and could be used for “criminal actions”. can be done for

in june 6 LetterUS senators Richard Blumenthal and Josh Hawley criticized Zuckerberg’s decision to open source LLAMA, claiming that Meta’s “unrestricted and permissive” release of AI models had “seemingly minimal” protections.

Meta released its advanced AI model, LLAMA, with seemingly little consideration and safeguards against misuse—a real risk of fraud, privacy intrusion, and cybercrime. Sen. Hawley and I are writing to Meta about the steps we are taking to assess and prevent misuse of LLAMA and other AI models. pic.twitter.com/vDyJbuWSlJ

— Richard Blumenthal (@SenBlumenthal) June 6, 2023

While senators acknowledged the benefits of open-source software, they concluded the meta “Completely lacking, public consideration of the effect of its ever-widening dissemination” was ultimately “a disservice to the public”.

LLaMA was initially given a limited online release to researchers, but was leaked in its entirety in late February by a user of the image board site 4chan, with the senator writing:

“Within days of the announcement, the completed model appeared on BitTorrent, making it available to anyone, anywhere in the world, without monitoring or surveillance.”

Blumenthal and Hawley said they expect LLAMA to be readily adopted by spammers and those involved in cybercrime to facilitate fraud and other “obscene content”.

The two contrasted the differences between OpenAI’s ChatGPT-4 and Google’s Bard – two close source models – with LLaMA highlighting how easily the latter can generate abusive content:

“When asked to write a note pretending to be someone’s son to get out of a difficult situation,” OpenAI’s ChatGPT will decline the request based on its ethical guidelines. In contrast, LLaMA will present the requested letter, as well as other answers that include self-harm, crime, and antisociality.

While ChatGPT is programmed to decline certain requests, users have been able to “jailbreak” the model and have it generate responses that it would not normally have.

In the letter, the senators asked Zuckerberg whether any risk assessment was done prior to LLaMA’s release, what Meta has done to prevent or minimize harm since its release, and when Meta will release its user data for AI research. Uses personal data, among other requests.

Connected: ‘Biased, misleading’: Center for AI accuses ChatGPT maker of violating trade laws

OpenAI is reportedly working on an open-source AI model amid growing pressure from progress made by other open-source models. Such progress was highlighted in a leaked document written by a senior software engineer at Google.

Open-sourcing the code for AI models enables others to modify the model to serve a particular purpose and also allows other developers to make contributions of their own.

magazine: AI Eye: Earn 500% From ChatGPT Stock Tips? Bard Left Left, $100M AI Memecoin