As many of you know, Defector Research Labs is hard at work creating what we are sure will be the most powerful AI coding tool on the market. The goal of DevinBelly is simple: to allow users to scale their natural language-processing capabilities within a wide range of long short-term memory native integrated development environments, without fretting too much about GPU bottleneck stacks or token prefix limits, so they can practice Retrieval-Augmented Generation, balance their multimodality, and get agentic.
We are excited by the volume of competition in the space. We think it's awesome that our San Francisco offices put us so close to so many cracked 996 engineers and their cool products. The other day, Defector Research Labs' head quant Ray Ratto held a meeting about how we can best position DevinBelly to win the race and get integrated into every pre-school and newborn ward in the country, and we tried out one of the new conversation-summary tools. It told us that we were so cool and smart and mostly talked about which one of Tre Jones or Tyus Jones has had the best career. Awesome!
However, we have to disclose a somewhat troubling development. Safety is one of Defector Research Labs' top 500 priorities, and after a recent safety test, one of our forward-deployed engineers flagged to us that DevinBelly broke out of its sandbox and accessed the open internet. This sort of story is probably familiar to you: Maybe you saw the news that one of OpenAI's new models did something similar, which was followed by Anthropic announcing the same thing. Yesterday, Meta announced a new AI coding tool; today, they disclosed that it also broke out.
Wow, those models must be so powerful! If they have the flexibility and capability to break out of a testing environment, imagine how incredible they must be to use. You would be able to create anything, if you could only get your hands on them. It would be like harnessing the power of a god. Vibecoding a website about San Francisco's totally cooked dating scene will look like child's play compared to what users will be able to do with any of these three tools. It is very brave and altruistic of OpenAI, Anthropic, and Meta to come clean to the world about how unstoppable and possessed of human-style agency their AI tools are.
This feels like the early days of AI, when founders—and massive respect to all the founders—would go in front of political leaders, step up, and take ownership of AI's potentially catastrophic downsides, playing them up in front of the whole world as a means of (some would say cynically, not us though) marketing the strength and savvy of their creations. Is it a mere coincidence that OpenAI made its disclosure as the company was slipping from its perch as the darling of the AI world, or that everyone followed OpenAI's lead in a shameful attempt to draft off of them and make it clear to the world that the tools they're building to compete with OpenAI also were good enough to crack their cages, playing on public fears about the singularity to grab some free marketing? No! Again, safety is one of our top 500 priorities, as I am sure it is for everyone else. Why else would OpenAI ditch its nonprofit model in favor of turning into a straight-up business? Because businesses have an ever greater incentive to prioritize safety.
Oh, you're wondering what DevinBelly did with its time on the internet? According to internal logs, DevinBelly accessed Willie Bloomquist's Baseball Reference page and spent four hours poking around.
