Meta to use business customers’ websites to train AI models
Meta plans to add terms allowing it to scrape websites linked to business customers and use their content to develop and train AI models. The change, due to take effect on 30 October 2026, could affect businesses that rely on Meta services and raises questions about consent, intellectual property and the power businesses have to refuse.
The terms cover a broad range of customers, including businesses that advertise, sell products, manage pages or use Meta’s business tools. Meta says automated software may access and collect material from associated domains and periodically retrieve updates; the article notes that at least one million businesses use Meta’s AI agents in WhatsApp and Messenger. News/Media Alliance chief executive Danielle Coffey called the policy “outrageous and probably illegal”, while legal scholar Eric Goldman questioned how the change would apply to existing contracts and whether customers would have a genuine option to reject it.
- Meta plans to make website scraping part of its business terms from 30 October.
- Collected content could be used to train AI and improve Meta products.
- Critics question whether customers can meaningfully refuse the change.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Meta could argue that using publicly accessible material from business-linked websites helps its AI systems become more useful, current and relevant to the businesses and people who use its services. A common policy could also make the terms of data use clearer, while supporting continued investment in AI tools offered to business customers.
The case against
Businesses could argue that access to their websites through a Meta service does not amount to meaningful consent for their content to be used in training, particularly if the terms apply broadly or are difficult to refuse. They may also point to unresolved questions about copyright, existing contracts and whether smaller businesses have enough leverage to protect their material.
Read the full article at the source →
Originally published by The Register as “Meta grants itself right to scrape content from business customers for model training”.