In order to respect consumers' privacy, Meta Platforms used public Facebook and Instagram postings to train its new Meta AI virtual assistant, but eliminated private messages shared only with family and friends, as per reports.

“Meta also did not use private chats on its messaging services as training data for the model and took steps to filter private details from public datasets used for training,” says Meta President of Global Affairs Nick Clegg.

“We've tried to exclude datasets that have a heavy preponderance of personal information. The vast majority of the data used by Meta for training was publicly available,” says Clegg.

Clegg's remarks come as tech companies such as Meta, OpenAI, and Alphabet's Google have been chastised for utilizing material scraped from the internet without permission to train their AI models, which consume enormous quantities of data to summarize information and generate visuals.

While facing lawsuits from authors accusing them of infringing copyrights, the corporations are debating how to handle the private or copyrighted materials swept up in that process that their AI systems may reproduce.

The most important product among the company's first consumer-facing AI technologies revealed by CEO Mark Zuckerberg was Meta AI.Connect, Meta's yearly product conference. Unlike previous conferences, which concentrated on augmented and virtual reality, this year's event was dominated by talk of artificial intelligence.

Meta created the assistant with a custom model based on the company's strong Llama 2 large language model, which was released for general commercial use in July.

Some image-generation tools enable the recreation of iconic characters such as Mickey Mouse, whilst others have paid for the resources or purposefully avoided included them in training data.