Press Release
Aerodrome voting opens May 28. Mainnet Launch: June 4. This quarter, AI started writing its own exploits. is shipping the trust layer underneath it. Code Is Abundant. Trust Is Not. In the span of seven days, the ground beneath the software shifted twice. On May 4, The Conversation published the most widely-circulated post-mortem yet of Anthropics Claude Mythos Preview, the frontier model Anthropic itself declined to release, because it can autonomously discover zero-days, generate working exploits, and execute multi-step cyber operations with minimal human oversight. Days later, Google‘s Gemma 4 landed inside Android’s AICore and Google AI Edge, putting agentic code generation, function calling, and offline reasoning on every developers phone and laptop under an Apache 2.0 license. The implication is unavoidable. When any device can generate, execute, and weaponize software autonomously, trust cannot live in the binary. It has to live at the source. Tea: the value layer for open source Tea is the provenance, attribution, and verification layer for a world where code is written by agents faster than humans can audit it. Every package, every contribution, every dependency, cryptographically attributed, continuously verified, and economically aligned with the people and systems that built it. Tea goes live on Aerodrom: the liquidity engine of base meets









