Not surprising, but still telling. Vishal Maini, who worked on Deepmind's communications and policy team from 2018 to 2022, says "external communication about the possibility of human extinction was not permitted, by anyone, at any level of the organization." Researchers were coached to dismiss those risks as alarmism, compare them to "movies like Terminator," and pivot to positive applications in healthcare or climate instead.

Internally, the team knew the AI alignment problem was not solved and "there were far too few people working on the problem." After months of pushing back, Deepmind loosened the rule. Positively framed safety content got the green light, including a blog post that wrapped extinction risk in friendly language. Maini says the gap between internal knowledge and public messaging is shrinking "because the evidence is harder to dismiss now."

His account comes as a growing number of AI safety researchers are speaking out publicly, warning about unsecured models that could enable dangerous actions or advanced systems that could slip beyond human control. Researchers at Deepmind itself are among those raising their voices.