Amazon employs an internal AI tool called Atlas to forecast unionization risks across its warehouses in the US, Canada, and the UK, according to documents disclosed during a Canadian labor dispute. Atlas aggregates "hundreds of data points," including social media activity (e.g., Reddit posts), internal feedback survey responses, workplace metrics, and historical labor actions, to assign warehouses risk scores between 100 (lowest risk) and 900 (highest risk). Warehouses are classified into five tiers, with Tier 1 indicating imminent union activity and Tier 5 signaling no immediate risk. Sixteen facilities globally are currently Tier 1, including two UK sites: a £500m robotic fulfillment center in Sutton Coldfield (EMA4) and the Coventry warehouse (BHX4), where workers previously staged strikes and narrowly lost a union recognition vote by 28 votes in 2024. Amazon denies individual employee monitoring, asserting Atlas operates at a "site/building level" to identify areas needing "support." However, legal representatives and unions allege the tool enables anti-union surveillance, citing past penalties such as a €32m GDPR fine from France’s CNIL in 2022 for excessive worker tracking. UK unions GMB and Foxglove are challenging Amazon’s use of "click-to-quit" QR codes and broader labor practices, while regulators emphasize transparency and workers’ rights to unionize. Amazon maintains it does not use Atlas in the UK but has not clarified whether similar systems exist under different names.