Toxic recommender algorithms: immunities, liabilities and the regulated self-regulation of the Digital Services Act and the Online Safety Act

With focus on toxic recommender algorithms, the article argues that the platform immunities granted in the early 2000s are incompatible with the regulated self-regulation of the EU Digital Services Act 2022 [DSA] and the UK Online Safety Act 2023 [OSA]. Whilst the DSA/OSA’s regimes recognise platforms as active shapers of the online content sphere, the immunities remain stuck in the early view of the same platforms as passive providers of neutral infrastructure and so as innocent messengers. Thus the immunities continue to shelter platforms from liability for harm caused by their toxic-but-profitable algorithms that the DSA/OSA seek to restrain. The tension between the two regimes may be resolved by restricting the immunities to truly neutral platforms, that is those not substantially invested in the content they are meant to regulate. For all other platforms the DSA/OSA self-regulatory regimes and standard liabilities could and should run in parallel.