Is an AI assistant safe? A checklist before you choose

aila team · · 4 min read

A safe AI assistant is one that does not act without your approval, identifies itself, handles your data under clear rules and states its fees up front. Ask a product the seven questions below, and look for the answers in its written privacy and terms documents.

Actions and control: three questions

An assistant is safe less because it speaks well and more because it writes its limits down. This list is not tied to any product; you can ask the same questions of whichever assistant you evaluate. If you cannot find an answer to one item, treat that as a signal too.

Start with what the assistant can do in your name.

  • Does it act without approval? Binding steps such as calls, quotes and bookings should depend on your approval. It is better if it shows what it will do beforehand.
  • Does it identify itself? If it speaks to someone on the phone, that person should know from the start that it is an AI.
  • Can you cancel? You should be able to stop a job you started and see the cancellation terms in advance.

Data and privacy: three questions

The second group is about your data.

  • How does it process personal data? Who the data controller is, who the data may be shared with and whether it is transferred abroad should be written in the privacy notice. A clear statement that data is not sold is a good sign.
  • Does it ask for explicit consent for sensitive data? Under KVKK, explicit consent is consent related to a specific subject, based on information and given with free will. Data such as voice, health and location should require it separately.
  • What happens to voice and recording data? Ask whether conversations are recorded, how long they are kept and who can access them.

Money and support: two questions

The last two items affect everyday use.

  • Are fees transparent? The limit of the free part and the price of separately paid services should be visible before you commit.
  • Can you reach a human? When something goes wrong, there should be a contact channel you can use.

How does aila answer these items?

The following are only behaviors stated in aila's published texts.

  • Approval: an approval card arrives before a call, and no call is made until you approve. Calls, quotes and bookings are never final without your approval. On topics you have not instructed, it does not decide on its own.
  • Self identification: when calling a business, it introduces itself as your AI assistant.
  • Data: the data controller is Humion Teknoloji Anonim Şirketi; data is processed under KVKK and is not sold to third parties for commercial purposes. It may be shared only as far as necessary with infrastructure and cloud providers, payment institutions, business partners and authorized public bodies; transfers abroad follow Article 9 of KVKK.
  • Explicit consent: voice recordings, health data and location information are processed only with your explicit consent. If you do not consent, that data is not processed, which may limit some features.
  • Fees: free within usage limits. Fees for drivers, transfers and couriers are separate, stated in the quote and paid to the provider.
  • Cancellation: you can cancel a job with Cancel on the task card. For drivers and couriers, cancellation terms are written in the quote.
  • Contact: for data requests write to info@myai.la; applications are answered within 30 days at the latest. The absolute accuracy of AI output is not guaranteed, so check important information yourself.

What you can do on your side

The November 2025 KVKK guide on generative AI also offers advice to individuals: choose carefully what you share with an assistant, avoid sharing identifying data such as identity and contact details and other people's information, and take care not to share sensitive information about health, finance and legal proceedings.

The guide also suggests reading the privacy notice and policy, finding out what data is collected, for what purpose, with whom it is shared and how long it is kept, and reviewing privacy settings. It is guidance, not legal advice.

Frequently asked questions

Check whether it acts without approval, whether it identifies itself, how it handles personal data and explicit consent, voice and recording data, fee transparency, cancellation and human support. Look for answers in its privacy and terms texts.

No. An approval card arrives before the call; calls, quotes and bookings are never final without your approval.

Only if you give explicit consent. Voice recordings, health data and location information are not processed without it.

No. The company does not sell personal data to third parties for commercial purposes, nor share it without permission.

It is safer not to share personal information that is not needed. The KVKK guide specifically advises avoiding sensitive information about health, finance and legal proceedings.

Sources

  1. KVKK: Generative AI and Personal Data Protection Guide (November 2025, Turkish) · Definition of explicit consent and advice to individuals on sharing data and reading privacy notices.