The news: Meta is going all in on open-source AI. The company has unveiled LLaMA 2, its first large language model that’s available for anyone to use—for free. It’s also releasing a version of the AI model that people can build into ChatGPT-style chatbots.
Why it matters: The idea is that by releasing the model into the wild and letting developers and companies tinker with it, Meta will learn important lessons about how to make its models safer, less biased, and more efficient.
But… Many caveats still remain. Meta is not releasing information about the data set that it used to train LLaMA 2, and it still spews offensive, harmful, and otherwise problematic language, just like rival models. Meta also cannot guarantee that it didn’t include copyrighted works or personal data, according to a company research paper shared exclusively with MIT Technology Review. Read the full story.
—Melissa Heikkilä
Spotting Chinese state media social accounts continues to be a challenge
It’s no secret that Chinese state-owned media are active on Western social platforms. But sometimes they take a covert approach and distance themselves from China, perhaps to reach more unsuspecting audiences.
Such operations have been found to target Chinese- and English-speaking users in the past. Now, a study published last week has discovered another network of Twitter accounts that seems to be obscuring its China ties—and this time, it’s made up of Spanish-language news accounts targeting Latin America. Read the full story.
—Zeyi Yang