AI Quick Wins for Safety Leaders

The rules catch up: HSE’s robot guidance and the new line on watching workers

The technology keeps arriving faster than the rulebook, and June proved it twice. In this episode, I work through two developments that close a bit of that gap. First, the HSE's announcement that it's writing Britain's first joint guidance on collaborative robots: why fear of non-compliance is holding firms back, and how cobot safety actually works, which comes down to the specific task it's doing. Then the data side: how safety wearables, location sensors and CCTV with computer vision all count as worker monitoring, what the Data Use and Access Act and the ICO's incoming code mean for you, and what to sort out before you switch any of it on.

The Right Tool for the Job: General-Purpose AI vs Specialist Safety AI

Most of us are already using tools like ChatGPT in some form, and over 900 million people use it every week. But when does it make sense to lean on a general-purpose AI assistant for your safety work, and when is it worth investing in purpose-built, safety-specific AI? In this episode, I share my perspective on where general AI assistants genuinely add value  and where they run into real limitations. I walk through three structural gaps that matter especially for health and safety: the lack of persistent memory between conversations, the challenge of keeping AI aligned with current regulations, and the risk of hallucinations presenting confidently incorrect information. I also cover the legal and GDPR landscape, including a recent UK High Court case involving fabricated AI-generated citations, the ICO's position on special category data, and the HSE's January confirmation that existing health and safety law already applies to AI in the workplace. I close with a look at what purpose-built platforms can offer, and a reminder that this isn't an either/or decision. It's about using the right tool for the right job.

What the International AI Safety Report 2026 Means for Health & Safety

The International AI Safety Report 2026 is the largest global collaboration on AI safety ever produced, backed by over 30 countries, authored by 100+ experts, and led by Turing Award winner Yoshua Bengio.  But what does a 200-page report actually mean for someone running health and safety? In this episode, I break down the key findings through a health and safety lens. You’ll hear about automation bias and the evidence that AI can quietly erode professional skills, why AI-generated outputs still can’t be trusted in high-stakes settings, the new threat of deepfakes to safety-critical communications, and a practical framework for adopting AI without losing your professional edge.  Whether your organisation is already using AI tools or just starting to explore them, this episode will help you ask the right questions.

From Copilot to Frontline Safety Tech at CES 2026

In this episode of Safety Rewired, we break down the biggest workplace safety signals from CES 2026, from hazard ready robots and in vehicle AI to exoskeletons, AR guidance and industrial digital twins. Expect a practical take on what’s deployable now, what’s still hype, and how to pilot tech without losing human oversight.

Why the Latest AI Models Actually Matter for Health and Safety

In just three weeks, three frontier AI models landed from three different companies. Google released Gemini 3. Anthropic shipped Claude Opus 4.5. And OpenAI pushed out ChatGPT 5.2 after an internal code red scramble. That compressed release cycle is not normal. It signals a turning point in how fast AI capability is moving and, more importantly, how work ready these tools have become. In this episode of Safety Rewired, Andy breaks down what this sudden acceleration in the AI arms race actually means for health and safety professionals. Not in abstract hype terms, but in the day to day reality of H&S work. You’ll hear a clear comparison of where Gemini, Claude, and ChatGPT now excel, why long context and reliability matter, and why ChatGPT 5.2 represents a step change for professional knowledge work. Andy also explores practical use cases already emerging in safety teams, from analysing large incident datasets to drafting policies and turning regulations into plain language. Just as importantly, the episode tackles the risks. Over trust, weak data, and governance gaps can turn AI from a productivity engine into a liability if it is not treated like a new worker on site. If you are responsible for safety and spend too much time buried in paperwork instead of on the floor, this episode will help you understand where AI can genuinely help right now and how to use it without losing professional judgement.