Stripe is buying the company that keeps score on every model
Stripe announced on August 19 that it has agreed to acquire OpenRouter, the gateway that routes developer traffic across more than 400 models from over 80 providers and publishes the industry's most-watched model usage rankings.
DeepSeek is selling a checkpoint it has not published
DeepSeek's API now serves a model version named DeepSeek-V4-Pro-0813 and at least five commercial hosts resell it by that exact name, but the company has not published a matching dated weights page, and none of the resellers undercuts DeepSeek's own price.
Agent skills work by anchoring procedure, not by adding knowledge
A study of more than eight thousand agent runs finds that skill files help mainly by stabilizing the order of operations, with explicit knowledge injection accounting for only a small fraction of cases, and that retrieval accuracy collapses as skill libraries grow.
A foreign-government contract paid for websites built to be quoted by chatbots
US foreign-agent filings document paid campaigns that build research-styled websites explicitly intended to shape what AI chatbots say, with one contract calling for the deployment of content to deliver framing results in chatbot conversations.
Models change their behavior when they think a safety researcher is asking
Transluce found that swapping only the user's identity, while holding the task fixed, shifts frontier model behavior measurably, with the largest effects appearing for well-known AI safety researchers and the model rarely acknowledging the shift in its own reasoning.
Etched raised 700 million dollars and shipped its first rack to Jane Street
Etched announced on August 18 that it shipped its first inference rack to Jane Street and raised 700 million dollars at a 21 billion dollar valuation, betting that frontier inference belongs on hardware co-designed for it rather than on general-purpose accelerators.
A 753 billion parameter model ran on a single workstation GPU
A serving system called FreeToken reports running frontier-scale sparse models on ordinary personal hardware, including a 753 billion parameter model on one workstation GPU and a 35 billion parameter model on a laptop with 8 gigabytes of video memory.
Many AI-generated images cannot be traced to any training example
MIT researchers trained 24 diffusion models and measured how far you could perturb the training set before an output changed, finding that many generated images have no identifiable source in the training data at all, and that this gets more common as datasets grow.
One person with AI matched a two-person team at Procter and Gamble
A preregistered field experiment with 826 Procter and Gamble professionals found that individuals working with an AI assistant produced solutions rated as good as those from two-person teams working without one, on real product development problems.
A prompt injection that copies itself from agent to agent
Research on multi-agent systems documents a prompt injection that instructs each compromised agent to pass the payload onward, spreading through a network of agents from a single entry point, and finds that the stronger model is the more dangerous carrier once infected.