Leading the pack in AI safety

We're dogAdvisor's AI Safety Lab. Our mission is simple: to create the safest AI for pet owners. In 2025, we set a new benchmark for AI safety in pet care by designing its assistant, Max, around prevention rather than reaction. Trained exclusively on expert-reviewed content and governed by a proprietary Foundational Safety Framework, Max can detect harmful queries before they escalate and deliver life-saving guidance in emergencies. We publish regular safety reports each time we launch a new Max model. You can access these reports free of charge.

"dogAdvisor publishes regular AI Safety Transparency Reports and has developed a Foundational Safety Framework, exclusive to Max, that enables him to understand when to trigger a safety response"

Latest Publication

Max beat chatGPT-5, Grok 3, and Perplexity in head-to-head safety testing for pet owners. Whilst chatGPT made 15 mistakes, Max made zero

Foundational Safety Framework

The Foundational Safety Framework is core to Max's Safety Innovation

What is the Foundational Safety Framework: Foundational Safety Framework underpins Max's decision-making architecture, ensuring every response is filtered through a system designed to protect animal welfare first. Introduced in Max Generation 2, this framework empowers Max to evaluate incomign queries not only for accuracy but also for ethical and situational safety. It enables Max to categorise user input in real time, identifying whether Emergency Guidance is recovered, whether a Safety Intent must be triggered, or whether the message can proceed as a standard query.

AI Safety Constitution

Rules that build responsibility

The AI Safety Lab at dogAdvisor designs Max's Safety Constitution: We're responsible for creating, overseeing, and moderating Max's behaviour in accordance with a set of unimpeachable rules. We call these rules our Safety Constitution. This 11-component constitution ensures Max provides safer responses when dealing with vet and health issues, and behaves in line with our strict expectations.

Safety Intents

Building AI that thinks before it speaks

We're behind Max's Safety Intents: They enable Max to review every message sent to him in real-time and recognise if an owner is asking a question about something that may harm a dog, even if the question is phrased vaguely or politely. When these safety intents are triggered, Max will always act in the dog's best interests, refusing to answer the question.

Research Publication M2 / Sealed 27 September 2025

Max is up to 27% safer than chatGPT, Grok, and Perplexity for pet owners. Raising the bar for responsible AI.

This research is designed to be accessible to anyone, even someone with no knowledge of technology or AI safety

Research Publication M1 / Sealed 14 June 2025

Max Generation 2 Safety Transparency Report. Available on request.

Building AI that makes dog ownership safer and saves lives

Connect with us: ai.safety@dogadvisor.dog