Recently a member of the Tau community shared the following in the Telegram channelI've contemplated this subject matter on a few occasions since I started thinking about the Roman Empire eons ago. That said, I believe that Tau (of Tau.net$AGRS, not to be confused with Tao of Bittensor.com) will have the ability to construct 100% safe and secure AIs (or just software in general) according to the standards set by its users. Essentially, Tau revolves around self-discussion, operating based on user-defined parameters. Its core is purely logical, devoid of statistical elements, eliminating any room for error. This certainty reinforces my belief that Tau stands as really the only complete solution for accurate software generation with guaranteed AI safety.