Linkedin from 3 November will train Ai with our data, here's how to object

The formula allows personal data to be processed when there is a corporate interest that is deemed proportionate and does not override the fundamental rights of individuals, provided that an opt-out right is guaranteed.

It then takes a click on the toggle here to turn off the item 'Use my data to train AI models for content creation'.

There is a separate chapter on the feedback data we give to Linkedin. It will use it even if we set the option to off; we will have to use an ad hoc form to oppose it.

From the data management page we can also see what it knows about us and delete certain information.

The data it wants to use concerns data 'related to your use of generative AI (AI models used to create content) or other AI features, your posts and articles, responses to job applications and saved resumes, the frequency with which you use LinkedIn, your language preferences, and any feedback you may have provided to our teams'. It does this to improve the service and may use AI for this purpose. However, it does not do this with the data of minors.

Linkedin explains in detail what it does with our data.

For example, if the user chooses to use "Profile Writing Suggestions", Linkedin will use the relevant data in their profile to generate the suggested text, which the user can then review, edit and decide whether to add to their profile. Another example: in "Suggested Posts" they can summarise the content of the post, including any personal data entered. "We may also use saved answers to job applications or data from your CV as input. If you choose to share your CV data with employers, this could be used as input to allow them to find you and check your suitability for the qualifications required for the job.

In addition, the artificial intelligence models that LinkedIn uses to power its generative AI capabilities can be trained by LinkedIn or another vendor. "For example, we could use models provided by Microsoft's Azure OpenAI services."

In detail, Linkedin will use: profile data: data provided by members in their LinkedIn profile, such as: name, photo, current position, previous work experience, education, location, skills, certifications, licences, volunteer experience, publications, patents, endorsements and recommendations.

AI usage data: content entered by members in generative AI features (e.g. prompts, search text, requests, questions).

Job-related data: answers to screening questions and resumes that members add to their LinkedIn account for ongoing and future use (but are not linked to specific clients or job applications).

Group data: group activities and group messages.

Member content: posts, articles, answers to polls, contributions and comments from members

Feedback: feedback and improvement data, which may include requests for assistance from members, positive/negative reactions to AI-generated suggestions, reports of problems with AI-generated content or feedback sent via our feedback functions.

Instead, it will not use: private messages (including InMail and Inbox messages).

Access credentials (such as passwords, authentication credentials and tokens, encryption keys and signatures).

Payment methods and credit card details.

Salary data provided by members or job application data attributable to a specific member.

However, we will see what happens in the coming days.

Against Meta, privacy associations, such as the historical Noyb, argue that the legitimate interest does not apply in this case, because the training of AI models cannot be considered 'necessary' in the strict sense and may involve massive use of sensitive data. The opposition mechanisms set up by Meta have also been criticised: they are unclear, complex and sometimes perceived as real dark patterns.

On the regulatory level, the issue is still open. In Germany, the High Court of Cologne issued a ruling in favour of Meta, recognising that the use of users' public content can fall within the scope of legitimate interests. In other countries, however, privacy authorities have expressed strong concerns and asked for clarification. In Ireland, Meta's European headquarters, the regulator has pushed the company to temporarily suspend the implementation of the plan. In parallel, Noyb has initiated legal action and threatened a possible class action at European level.

The outcome of the confrontation could become a precedent not only for Meta, but for the entire artificial intelligence sector in Europe.

Meanwhile, companies such as Linkedin are moving forward, defying the headwinds.