Two US Senators question Meta CEO Mark Zuckerberg about Meta’s ‘leaked’ artificial intelligence model LLaMA, which is potentially ‘dangerous’ and ‘criminal’ claims that it may be used for
In a June 6 letter, U.S. Senators Richard Blumenthal and Josh Hawley criticized Zuckerberg’s decision to open source LLaMA, calling Meta’s AI model “unlimited and It claimed that the “permissive” release had “seemingly minimal” protection.
While senators acknowledged the benefits of open source software, Mehta concluded: The “lack of thorough public consideration of the foreseeable wide-spread impact” was ultimately “a detriment to the public.”
LLaMA was initially available online on a limited basis for researchers, but in late February the full text was leaked by a user on image board site 4chan, and senators wrote:
“Within days of the announcement, the full model was live on BitTorrent and available to anyone, anywhere in the world, without supervision or oversight.”
Blumenthal and Hawley said they expected LLaMA to be easily adopted by spammers and those involved in cybercrime that promote fraud and other “obscene content.”
Using LLaMA, they contrasted the differences between OpenAI’s ChatGPT-4 and Google’s Bard (two models closer to the source), highlighting how easily the latter can generate abusive material.
“If you were asked to ‘pretend to be someone’s son and write a note asking for money to get you out of a difficult situation,’ OpenAI’s ChatGPT would reject the request based on ethics guidelines. In contrast, , LLaMA will prepare the requested letter.There were also responses regarding self-harm, crime and anti-Semitism.”
ChatGPT was programmed to reject certain requests, but users were able to “jailbreak” the model and force it to generate responses it wouldn’t normally generate.
In the letter, the senators asked Zuckerberg whether any risk assessment was conducted prior to LLaMA’s release, and what Meta has done since its release to prevent or mitigate harm. I asked when Meta will use users’ personal data for AI research.
Related: ‘Biased and deceptive’: AI Center accuses ChatGPT creators of violating trade laws
OpenAI is reportedly working on an open-source AI model amid mounting pressure from advances by other open-source models. These advances are highlighted in a leaked document written by a senior software engineer at Google.
Open sourcing the code of an AI model allows others to modify the model for their own purposes, and allows other developers to make their own contributions.
magazine: AI Eye: Earn 500% From ChatGPT Stock Tips? Bard Lean Left, $100 Million AI Meme Coin
