We do not need another slide deck on whether artificial intelligence will matter; we need regulators who can answer the same question the AI LIVE conference will pose to two thousand executives on 20 October: what are you allowed to deploy in production on Monday morning?
Enterprise speed, statute lag
The Olympia lineup includes Met Office chiefs, Vodafone security leads and retail automation heads—organisations already running models on customer data, supply-chain telemetry and HR records. Their panels focus on agentic workflows, cyber defences and drug-discovery pipelines, not on waiting for an Act of Parliament. That is rational behaviour; it is also how shadow IT becomes shadow policy when ministers split AI safety, digital competition and sectoral enforcement across multiple bodies.
Westminster's pro-innovation white paper promised context-specific rules enforced by existing regulators. In practice, the Information Commissioner's Office, Financial Conduct Authority and Ofcom are issuing overlapping consultations faster than chief compliance officers can harmonise them. Employers face ICO record-keeping duties on generative tools while the FCA asks whether model explainability satisfies Consumer Duty on credit promotions. None of those frameworks were designed together.
What good would look like
We are not calling for a moratorium. We are calling for a single public calendar that tells a hospital, bank and retailer which tests are mandatory before launch, with worked examples rather than principles. DSIT should publish cross-regulator playbooks tied to real incidents—phishing copilots, biased hiring screens, trading-desk summarisers—not aspirational ethics essays.
Until then, events like AI LIVE will keep selling tickets on the gap itself. Enterprise buyers will learn from each other's war stories while MPs debate water nationalisation. The government should treat that asymmetry as a competitiveness risk, not a talking point for innovation week.
Our view
Burnham's team wants Britain to lead on safe AI; leading means predictable rules enterprises can implement without hiring a consultancy for every feature flag. The speaker slate at Olympia proves demand is here. Westminster's job is to catch up before the next enforcement headline makes "British AI" a cautionary tale rather than an export story.
Parliament's calendar problem
Select committees can hold evidence sessions on AI, but they cannot issue binding technical standards. The result is a patchwork of voluntary codes that vendors cite in sales decks while public servants wait for Treasury approval to buy licences. We need Cabinet Office-chaired timelines that align ICO enforcement dates with FCA handbook updates so compliance teams are not whipsawed.
Until that happens, conferences will continue to substitute for statute—and taxpayers will fund both the booths and the breach clean-ups.
Ministers should attend Olympia not to cut ribbons but to hear which deployments already failed internal audit—and write rules from those failures.