The Seqlense Regulatory Brief - Edition #28 - EN

Edition #28 documents supervision entering the post-deployment phase: regulators are no longer looking only at training data or a model’s compliance at launch, but at its actual behaviour. On 9 October 2026, the European Commission convened a special meeting of the AI Act’s scientific panel to examine recent “loss of control” incidents involving frontier models, a first public mobilisation of the Articles 55, 68 and 90 framework, without any infringement being established. In the United Kingdom, the ICO secured practice changes from ten foundation model providers and opened a call for evidence on AI agents, while the government accepted 44 recommendations introducing whole-life-cycle monitoring of medical AI devices. The weak signal: the object of supervision is shifting from the model to the action.