Meanwhile Wikipedia opens up to more AI innovation
Meta, the company behind Facebook, is to use the conversations users have with its AI assistant – both voice and text – to provide even-more personalised advertising.
The policy will be implemented worldwide from 16th December and could affect up to one billion people worldwide, although users in the UK, European Union and South Korea will be excluded on data privacy and protection grounds. The aim, presumably, is to start driving revenue from AI, which is an expensive service to provide given the IT overhead required to support it.
“If you chat with Meta AI about hiking, we may learn that you’re interested in hiking – just as we would if you posted a reel about hiking or liked a hiking-related page. As a result, you might start seeing recommendations for hiking groups, posts from friends about trails, or ads for hiking boots,” the company stated in an announcement revealing the move.
Affected users will be notified on 7th October.
The data extracted from Meta AI conversations will be added to its existing targeted advertising database that contains information about people’s apparent likes and dislikes. However, the company claims that will not deliver targeted advertising based on sensitive topics, such as sexual orientation, political views, health, racial or ethnic origin.
The shift is consistent with founder and CEO Mark Zuckerberg’s aim to make “Meta AI the leading personal AI with an emphasis on personalization, voice conversations and entertainment”, a goal stated at the company’s annual general meeting earlier this year.
But it’s open to question whether Meta AI can currently compete with ChatGPT or Grok. When we asked Meta AI to provide more information about this initiative, it was unable to provide any sources from the company, but it did add that “users won’t have an option to opt out of this data collection, except by not using Meta’s AI tools”.
Meta is not the first company to start deploying AI in the service of ad personalisation. Google and Amazon have already made a start via their cloud-based services, but Meta is the first to enlist users AI conversations in the service of driving advertising revenues.
Wikipedia’s AI moves
At the same time, Wikipedia has formally launched a new project that could see the open-source encyclopaedia open up to AI models.
Called the Wikidata Embedding Project, it is being led by Wikimedia Deutschland in collaboration with Jina.AI and DataStax. “The project seeks to support the open source community in developing innovative AI applications, using Wikidata’s multilingual and inclusive knowledge graph, while making its extensive data more accessible and contextually relevant for users across the globe,” according to Wikidata.
The project started in 2024, but only officially launched on 1st October 2025.
The organisation has offered machine-readable data from its properties for a number of years, but the information could only be exploited in the form of keyword matching, which can miss the real meaning behind a search query.
“SPARQL [Wikidata’s Query Service] offers precise and accurate data retrieval, but its steep learning curve and complexity make it challenging for many users to leverage. The vector-based [semantic search] approach bridges this gap, combining the accessibility of keyword searches with context-aware results,” the organisation explained.
The database is now accessible on open-source website Toolforge.