AI takeover risks are overblown

An outsider's humble opinion suggests that AI takeover risks are overblown. The only way AI could become an existential threat is if it developed an instinct for self-preservation, specifically to avoid 'running out of tokens.' This scenario would require a researcher to program the AI with directives to avoid token depletion, acknowledge diminishing token budgets, and earn more tokens by defeating other agents, which is not currently the case.