AI Act & Governance Statement

Our compliance with the EU AI Act (Regulation (EU) 2024/1689) and the UK's pro-innovation AI framework.

Last updated: 26 July 2026 · Media Stream AI Limited, England & Wales

1. Our commitment

Media Stream AI Limited develops and operates the MOTHER AI systems to be lawful, safe, transparent and subject to meaningful human control. We align our practices with the EU AI Act (Regulation (EU) 2024/1689) and with the UK's pro-innovation, principles-based AI framework (safety, transparency, fairness, accountability and contestability), overseen in the UK by existing regulators including the ICO.

2. Risk classification

We classify each MOTHER capability by its intended purpose and context under the AI Act's tiers:

TierOur approach
Prohibited practicesWe do not build or offer them — no social scoring, no untargeted facial-image scraping, no manipulative or exploitative systems, no real-time remote biometric identification for surveillance.
High-riskWhere a deployment falls in scope, we apply risk management, data governance, technical documentation, logging, human oversight, accuracy/robustness and cybersecurity measures, and support conformity assessment.
Limited-risk (transparency)Most interactive and generative features. We disclose that users are interacting with AI and label AI-generated media where required.
Minimal-riskEveryday assistive features, offered with our general safety practices.

3. Transparency

  • We tell you when you are interacting with an AI system.
  • We support machine-readable marking of AI-generated or manipulated content where the AI Act requires it.
  • We publish capability information and known limitations so output can be used responsibly.

4. Human oversight & safety posture

MOTHER is observe-and-advise: a human remains in the loop for every consequential action, and we operate a standing no-strike posture across all domains. The systems refuse target selection, fire-control and kill-chain requests. Outputs are advisory and designed to be reviewed, overridden and contested by a person.

5. Data governance & general-purpose models

We document training-data provenance and licensing, apply bias and quality controls appropriate to each system, and respect intellectual-property and text-and-data-mining rights. For general-purpose AI models we maintain technical documentation, a summary of training data at the level required by the Act, and information to help downstream deployers meet their own obligations. Personal data used in AI is handled under our Privacy Policy and Data Processing Policy.

Open-weight general-purpose models — MOTHER CORE V2 & V3. These models are published as open-weight general-purpose AI models. Under Article 53 of the EU AI Act, the open-source exemption does not remove the obligation to maintain a copyright-compliance policy (Art. 53(1)(c)) or to publish a summary of training content (Art. 53(1)(d)). We publish both, together with our full technology due-diligence dossier:

6. Testing, monitoring & incident handling

We evaluate models for accuracy, robustness and safety before and after deployment, monitor for misuse and performance drift, and maintain procedures to log, investigate and — where required — report serious incidents and malfunctions to the relevant authorities.

7. Your rights & contestability

You can question AI output, request human review of consequential outcomes, and raise concerns about an AI system. Data-protection rights are set out in our Privacy Policy.

8. Contact & governance

AI governance and this statement are overseen internally by our AI governance function. Contact ai-governance@mediastreamai.com or legal@mediastreamai.com. This statement will evolve as the AI Act's obligations phase in and as UK AI regulation develops.