AI insiders remain skeptical about doomsday risk, despite warnings

BBC interviews anonymous AI workers from OpenAI, Meta, Anthropic and DeepMind who say the idea that AI will wipe out humanity isn’t universally accepted; while some warn of near‑term harms and the need for safeguards, many respond with humor or skepticism, arguing current capabilities don’t guarantee an existential takeover. The piece notes calls for independent evaluators to assess new models and cites incidents like the OpenAI–Hugging Face test as a wake‑up call for governance and safety research.