Investors — and impersonators — are converging on the AI space. Gemma Postlethwaite, chief executive officer of expert network GLG, told Semafor that anyone “who knows anything about AI” is in high demand from investors looking to play the AI trade. That demand, she said, has increased the risk of deepfakes and of people using AI chatbots to pose as experts.
Postlethwaite said GLG must ensure that experts are not only who they claim to be but also genuinely knowledgeable about their stated fields. To do that, GLG has introduced additional screening and identity verification, increasingly conducted on camera with questions specifically designed to trip up AI avatars.
Verification techniques
GLG also parses interview transcripts to detect occasions where a person’s claimed expertise starts to go off‑piste. Postlethwaite described this approach as building a kind of "digital fingerprint of knowledge": verify first, then trust.
Connecting domain experts with AI labs
In addition to authentication measures, GLG is connecting frontier AI labs with subject‑matter experts to help train models on sensitive topics. Examples include child psychiatrists who can help craft responses for children in crisis, and nuclear or bioweapon specialists whose expertise can be used to counteract bad actors seeking operational instructions.
GLG’s actions reflect efforts to respond to rising, more complex risks driven by strong market demand for AI expertise.
(Text: Liz Hoffman — source: interview with Semafor)
