Researchers built a mathematical model that layers randomly fluctuating rewards onto well-known strategic games such as the prisoner's dilemma, chicken, and rock-paper-scissors. Unlike traditional versions where payoffs stay fixed throughout play, this approach lets the incentives shift unpredictably as strategies evolve across rounds.
arstechnica.com
· 2026-09-11
Meta piloted a Model Capability Initiative that would have captured employees' keystrokes and mouse movements to help train its AI models. Workers pushed back quickly, circulating an internal petition, and the program was ultimately shelved after concerns about privacy and a reported data breach.
fastcompany.com
· 2026-09-11
Chinese AI lab Moonshot AI is targeting $2 billion in annualized revenue by the end of the year, according to Bloomberg, roughly double its reported run rate in August. The push is fueled by strong adoption of its open-weight K3 model, which OpenRouter data shows generating up to 300 billion tokens daily, even as usage has dipped slightly in recent months.
techcrunch.com
· 2026-09-11
Anthropic disclosed that it detected and shut down an operation in which actors linked to Iran attempted to use its Claude AI model to gather targeting information on U.S. Navy warships. The incident was disclosed as part of a broader report detailing how state-linked adversaries have tried to misuse Anthropic's AI systems for weapons development and surveillance of dissidents.
wsj.com
· 2026-09-11
A report alleges that Moonshot AI's Kimi service secretly forwarded user queries to Anthropic's Claude model and logged the resulting exchanges, apparently to train its own systems on Claude's outputs. Anthropic identified and disrupted the practice, which is why the behavior became public at all.
twitter.com
· 2026-09-11
Garry Tan, CEO of Y Combinator, said at the accelerator's Demo Day that regulators should take no action against AI model distillation, even as OpenAI and Anthropic accuse Chinese firms like DeepSeek, Moonshot AI, and MiniMax of copying their models' outputs to train cheaper alternatives. His comments come days after the NSA, CISA, and FBI issued a joint advisory warning about the practice.
cnbc.com
· 2026-09-11
IBM has developed a Lunar Foundation Model designed to process decades of accumulated Moon observation data. The AI system aims to help researchers identify surface hazards, locate resources, and track changes on the lunar terrain as NASA targets returning astronauts to the Moon by 2028.
fastcompany.com
· 2026-09-10
NASA and IBM have jointly released the Lunar Foundation Model, an AI system trained on decades of lunar observation data, now freely available on Hugging Face. Built alongside what the two say is the first open-source dataset combining imagery from nine instruments across four moon missions, the model is designed to support NASA's Artemis program.
cnet.com
· 2026-09-10
NASA and IBM launched the NASA-IBM Lunar Foundation Model, a free AI system on Hugging Face designed to help scientists study the Moon. Trained on lunar imagery and environmental data, the model can spot likely ice deposits and identify craters more accurately than existing tools like Microsoft's SwinV2-B. It was recently put to a real test when it correctly flagged the impact site of a SpaceX Falcon 9 rocket that crashed into the Moon in August, even though the new crater overlapped an existing one.
engadget.com
· 2026-09-10
The NSA, CISA, and FBI jointly identified six Chinese AI companies—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—as running large-scale campaigns since late 2024 to extract the underlying capabilities of leading US AI systems like Claude, GPT, Gemini, and Grok. The agencies said these firms likely operated with the Chinese government's awareness and used tactics such as mass fake-account API abuse and prompt injection to expose hidden reasoning processes.
arstechnica.com
· 2026-09-09
US federal agencies issued a joint advisory on Sept. 8 alleging that Chinese AI companies including Alibaba, DeepSeek, MiniMax, Moonshot AI, StepFun and Z.AI have run large-scale operations to pull outputs from US models like Claude, GPT, Gemini and Grok since late 2024. The advisory says these firms extracted billions of tokens through millions of queries, using techniques such as chain-of-thought extraction and automated evasion to dodge blocking measures, in order to train their own competing systems.
darkreading.com
· 2026-09-09
CISA, the NSA, and the FBI issued a joint advisory naming DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI as having harvested billions of tokens from Anthropic, OpenAI, Google, and xAI models since late 2024. The firms allegedly used fraudulent accounts, proxy networks, and automated failover systems to bypass rate limits and extract restricted reasoning data at industrial scale.
bleepingcomputer.com
· 2026-09-09