London-based research engineer Sofia Liguori, who had worked at the company for three years, resigned that week; researcher Andreas Kirsch resigned earlier in the month.
According to the report, Google signed an agreement earlier in 2026 allowing its AI models to operate in classified Pentagon environments. The agreement permits use for “any lawful government purpose,” while restricting domestic mass surveillance and autonomous weapons, including target selection, without appropriate human oversight and control.
Liguori told the BBC she had lost confidence in Google’s ethical commitments and alleged staff feared retaliation for public criticism. Google said it has reporting channels and anti-retaliation policies. Kirsch questioned Google’s ability to see or veto Pentagon uses of its models. The report does not establish that Google models independently select battlefield targets, the full terms of the agreement, or how many employees have resigned over these contracts.