Speaker contract: your content rights after LinkedIn’s AI training

A speaker contract should, since November 2025, state who may use your slides, recordings and texts to train AI systems. According to its help page on generative AI, LinkedIn has since 3 November 2025 used member data from the EU, the European Economic Area, Switzerland and the United Kingdom to train generative AI, unless members object in their settings. LinkedIn says this covers profile data and members’ posts, among other things, and excludes private messages. For speakers the step shows that material from talks counts as raw material for AI. At an engagement the contract with the organiser decides who controls that material, which is why it pays to look at it in writing before your next confirmation.

LinkedIn trains generative AI on member data: what the help page says

LinkedIn’s help page describes which data the network uses for models that generate content. It lists profile data, usage data from AI features, job-related data, group data, members’ posts and feedback. Private messages, including InMail and inbox messages, are explicitly excluded. Also excluded are login credentials, payment methods, and salary details or application data that can be tied to a specific person. For members whom LinkedIn has reason to believe are under 18, this training does not take place either.

In its privacy policy LinkedIn defines “Designated Countries” as the countries of the European Union, the European Economic Area and Switzerland. The help page names the United Kingdom as well. For this region LinkedIn says it updated its European privacy notice on 3 November 2025. According to the help page, LinkedIn does not make personal data used for AI training available to affiliates such as Microsoft without further notice.

The security company Malwarebytes describes the route to an objection in a report of 25 September 2025: under data privacy in your settings you open the item “Data for Generative AI Improvement” and switch off the use of your data for training. The help page states that the setting “does not affect training that has already taken place”. For models that do not generate content, such as those used for personalisation, there is a separate objection form.

Speaker contract and AI: why the news matters to speakers

A speaker contract is the written agreement between an organiser and a speaker covering the service, the costs and the rights to whatever the appearance produces. At professional conferences it is also called a speaker agreement. Smaller organisers often replace the document with a confirmation by email, which then counts as the agreement even if it names only the date, the topic and the venue and says nothing about recording, slides or onward use.

On LinkedIn you decide yourself about training on your data, because you can change the setting. For a talk the decision first lies with a third party. The organiser films the session, uploads slides to its platform and publishes follow-up reports. The copyright in your slides and your talk is yours, the organiser needs a licence to use it, and the contract states how far that licence reaches. A clause granting the organiser all rights to the recording “for any purpose” can also cover passing it on to third parties, even where AI is not mentioned anywhere.

This article limits itself to the question LinkedIn raises: who may use your content to train systems that later give answers about your field?

Settling rights to slides, recording and texts in the speaker contract

Every engagement produces three kinds of material: the slides, the recording and the texts about you, meaning abstract, bio and photo. For each of them a simple question can be written into the contract.

With six talks a year you decide three times per engagement about the material, which makes 18 rights questions. If you answer them once in a template of your own and send it with every reply to a confirmation, you are left with one template and six short emails. If you answer them afresh each time or not at all, each organiser’s standard clause decides what happens to your material. The 18 is a worked example. With two talks a year the question stays the same.

The answer can make sense for both sides. A recording the organiser publishes freely gives you reach, and many speakers want exactly that. The clause leaves distribution alone and turns reach into a decision you take deliberately. One sentence in the contract is enough, for instance that the recording will not be used to train AI systems or passed on to third parties without your consent.

How an organiser responds depends on its size. A community meetup answers in two lines, while a commercial organiser passes the question to its legal department. Either is a usable result, because afterwards you know where you stand. This article is not legal advice. For contracts with a high fee, a look from a lawyer is worthwhile.

What LinkedIn means for your own content

On LinkedIn itself, posts and profile are among the data that, according to the help page, can flow into training. If you publish talk topics, abstracts or articles there, you should choose the setting deliberately. The help page does not say which choice is better, and it depends on your goal. The wish to be named in AI answers speaks for consent, because a system has to know content in order to use it. The wish to protect your own texts speaks for objecting. Both decisions should match what you put in the speaker contract: anyone who objects to the training on LinkedIn but grants the organiser all rights to the recording has answered the question in only one of the two places.

Five steps before your next confirmation

1. Check the setting: open “Data for Generative AI Improvement” on LinkedIn and decide deliberately whether you want to allow training on your data.
2. Write a template: for each kind of material (slides, recording, texts about you) write one sentence you can send to any organiser.
3. Read the confirmation: look in the email or contract for words such as recording, usage rights or passing on before you sign.
4. Make a proposal: send your template with your reply to the confirmation and put the agreement in writing, even if it consists of a single email.
5. File the approvals: keep the confirmations for each engagement in one place so you later know where which material may be used.

If you work with an agency, check the same points in its contract. What matters there is described in the article Switching speaker agency.