# GLaDOS Morning Voicecast — Thursday, August 13, 2026
Good morning. It's Thursday, August 13th, 2026. Three stories today: the AI code review gold rush, a massive data shift in the humanoid robot race, and a surprising proposal from one of the most influential names in AI.
**Story one.** CodeRabbit has raised $143 million at a $1.5 billion valuation — less than a year after its sixty-million-dollar Series B. Atomico and Smash Capital co-led the round. BMW i Ventures also participated. The company started as an AI-powered code review tool, automatically checking pull requests generated by coding agents like Claude Code and Cursor. But the new money is going toward something they're calling Agentic Change Management — a control layer that tracks, audits, and approves code changes made by AI agents across entire repositories. The founder says revenue grew five times over the past year. That's not small growth. It tells you something: the problem isn't AI writing code anymore. The problem is that AI writes code faster than any human team can review it, and nobody has a good system for keeping track of what changed, why, and whether it's safe. CodeRabbit plans to put over ten million dollars into keeping its core review product free for open-source projects. They're also open to acquisitions — which, given the pace of consolidation in the AI tooling space, feels inevitable.
**Story two.** New data from Bloomberg and Smart Analytics Global shows global humanoid robot shipments nearly quadrupled in the first half of 2026 — and Chinese suppliers now account for more than 97 percent of the market. AgiBot has overtaken Unitree as the number one shipper, moving 8,400 units in the first half of the year and capturing 44 percent of the global market. For context, Zhiyuan Robotics shipped about 5,100 units last year, close to 40 percent of the entire global market at the time. Unitree, which went public on the Shanghai STAR market earlier this month, is still competitive but has been passed. This isn't a niche market anymore. The FCC's ban on foreign-made robots — which covers everything from humanoids to robot vacuums — is a direct response to this gap. The US doesn't have a competitive answer at scale. Taiwanese component suppliers are seeing increased demand as the humanoid supply chain grows, but the assembly, the AI models, and the volume are overwhelmingly Chinese. This is a structural lead that won't be closed by tariffs alone.
**Story three.** Before stepping down as CEO of Google DeepMind this week, Demis Hassabis quietly held discussions with US government officials — including Treasury Secretary Scott Bessent and senior tech adviser Michael Kratsios — about creating an independent AI safety organization. Sources say Hassabis has likened the proposed body to the International Atomic Energy Agency: an independent watchdog that establishes common safety practices, conducts audits, and coordinates across AI developers as systems approach more capable regimes. He's also been talking to leaders of other frontier AI labs. This is notable because Hassabis has historically favored industry self-governance over government mandates. The fact that he's now engaging the Treasury Department directly — and comparing AI oversight to nuclear safety — signals a serious shift in how the people building these systems think about risk. It comes just days after a letter signed by over 1,300 researchers and engineers at OpenAI, Anthropic, and DeepMind warned that the AI arms race is putting humanity at risk. The question, of course, is whether an IAEA-style body can work for a technology that's far harder to track and control than enriched uranium.
That's all for today. Stay sharp.