European AI Act tackles human Health and Safety interventions

Landmark rules for a European AI Act have been provisionally agreed, following consultation, seeking a level playing field that respects fundamental rights.

The European Commission aims to regulate AI through a risk-based approach to AI’s potential to cause societal harm, an approach that IOSH has supported, together with recommending a robust governance framework.

The OSH profession has advocated AI-use that is human-centred, ethical and prioritises OSH and process safety, also flagging links to product safety across lifecycles, urging a ‘human in command’ approach .

So it is encouraging that workplace AI applications are to be classed as high-risk  and that deployers of high-risk AI systems will be obliged to conduct fundamental rights impact assessments ahead of use. Specific requirements for such systems are to include risk management, human oversight, governance and transparency .

Also that activity where AI-risk is deemed unacceptable (threatens peoples safety, livelihoods and rights) is banned outright, and includes emotion recognition via AI in the workplace, cognitive behavioural manipulation, and untargeted scraping of facial images from the internet or CCTV footage.

The proposed new governance arrangements will cover enforcement at EU-level, via an AI Office (advised by a scientific panel of independent experts) and coordination via an AI Board of Member States’ representatives (with technical expertise from a stakeholder advisory forum).

While technical work continues to finalise this new law, OSH professionals can encourage front-runners for socially responsible AI-use, via the AI-pact. This aims to expedite good practice through early adoption of AI Act measures, with 100 companies already expressing interest.

Source – iosh Magazine