AI Act and your communication

AI in AAC: meaning, control and transparency

AI can suggest how a message is worded and help prepare symbols and board drafts. The reference point is the meaning chosen by the person who uses AAC. Every feature needs a clear description of how it works and control adapted to the person.

The AI Act is the European Union regulation on artificial intelligence. On this page we explain the role of AI in the app, the principles for using suggestions and how to get support.

1. Your message and your authorship

The words, symbols and intention chosen by the person are the starting point. You compare an AI suggestion with your own message and choose the version you want to share. You can keep your own message, or accept the suggestion and keep changing it.

2. Clear information about AI involvement

Suggestions, symbols and board drafts prepared by AI are presented as a proposal — nothing replaces your message unless you choose it. How proposals look on each device, which AI features are on by default and how to switch them off is described in the AI transparency notice.

3. Control the person can use

How suggestions are checked is adapted to the person’s reading, understanding, vision, hearing and device use. Accepting or rejecting a suggestion should be possible with the same access method the person uses every day. The partner’s support is agreed individually, respecting the person’s decisions.

4. Core communication on the device

Boards, locally available symbols and synthetic speech are the core way of communicating. They work offline. You switch online AI features on and off in settings, and the Simple profile hides sentence proposals — you can go back to core AAC at any time.

5. Accuracy and preserving meaning

AI produces suggestions that may need correcting. Particular care is needed with refusals, negation in the user’s message, who is speaking, when something happened, pain, urgent needs, and unusual or short messages. Checking covers what the message means and whether it matches the person’s choice.

6. Stability and safety

Core communication stays available when there are problems with the connection or the AI service. Information on supported hardware, languages, features and limitations is next to the relevant settings and in the manual.

7. Data and informed choice

Before an online feature starts, you are told what data is sent, why it is processed and which service providers are involved. You can review and change individual consents and settings in the Privacy Center.

8. Vocabulary and developing communication

Vocabulary choices and board changes follow the person’s needs and what is agreed together with communication partners and the AAC professional. AI suggestions are material to check and adapt.

9. Support and reporting

If an AI suggestion changed the meaning of a message, describe the words or symbols you chose, the suggestion you received and the meaning you expected. Include the information needed to explain what happened. The support team will help you work out the next steps.

The AI Act — for those who want the details

Article 50 covers transparency for certain uses of AI, and Article 4 covers AI literacy. Article 14 describes human oversight, and Article 15 accuracy, robustness and cybersecurity of high-risk systems. These principles are also a valuable reference point when designing communication support. Which obligations apply to a specific feature depends on its intended purpose and classification.