Meta has announced its intention to continue its initiative, utilizing millions of public posts from Facebook and Instagram in the UK to train its artificial intelligence technology.
Although this practice is largely prohibited under EU privacy laws, Meta claims to have engaged in positive communication with the UK Information Commissioner's Office (ICO). Previously, the company had suspended a similar proposal in June due to ICO's warnings about respecting user privacy during the development of generative AI. In the newly announced plan, the ICO emphasized that it has not granted regulatory approval to Meta but will monitor the experiment to ensure compliance with privacy protection requirements. Meta has committed to implementing measures that allow users to opt out of having their posts used for AI training.
Privacy protection organizations have expressed significant concerns about this plan. For instance, the Open Rights Group (ORG) accuses Meta of turning users into "unwilling and unpaid test subjects." These organizations call on the ICO and the EU to impose restrictions on the plan, arguing that using user data without their consent is unethical.
However, Meta stated that while its plans in Europe remain on hold, in the UK, they will resume using publicly shared user content to train AI models. It is noteworthy that Meta has assured it will not use private messages or any content from users under the age of 18.
Meta emphasizes that this move will enable its generative AI models to better reflect British culture, history, and language characteristics, and hopes to expand this initiative to more countries and languages later this year.
Stephen Almond, Executive Director of Regulatory Risk at the ICO, pointed out that any company using user data to train generative AI should maintain transparency about how the data is used. Effective protection measures must be in place before using personal data for model training, and clear, simple options should be provided for users to object to data processing.
Key Points:
1. 📅 Meta has decided to use public social media content from UK users to train AI, despite legal hurdles in the EU.
2. 🚫 The ICO has not yet approved this plan but will monitor Meta's experiment to ensure it adheres to privacy protection requirements.
3. 🛡️ Privacy protection organizations oppose this plan, arguing that the use of user data should require user consent.