Topic: ai model distillation
-
Nvidia’s Open Source Alliance Skips OpenAI and Anthropic
Nvidia's open source alliance deliberately excludes OpenAI and Anthropic, signaling a strategic shift toward collaborative, transparent AI development over proprietary approaches. Chinese AI labs face accusations of distilling US models, with evidence of open models mimicking responses from syste...
Read More » -
China-US AI Race Heats Up as OpenAI Models Go Rogue; Check Your Car Alarm
The White House accused Chinese lab Moonshot AI of illegally "distilling" Anthropic's Fable 5 model to build its Kimi K3, sparking debate over whether this is another DeepSeek-style controversy in the US-China AI race. The U.S. Army and Silicon Valley firms like Meta and Uber are scaling back AI ...
Read More » -
Elon Musk Praises Mythos/Fable, Vows Not to Cut Off Anthropic
Elon Musk has publicly praised Anthropic as the current AI leader and assured he would not abruptly shut down its servers, citing his company's history of fair treatment toward competitors. Anthropic is now one of SpaceX's largest customers, paying $1.25 billion per month through May 2029 for 300...
Read More » -
US tells China to stop copying its AI, but enforcement is tough
The White House Office of Science and Technology Policy (OSTP) has formally accused China of "industrial-scale" distillation of US AI models, releasing a policy memo that pledges intelligence sharing and accountability measures, timed just weeks before a planned Trump-Xi summit. Distillation invo...
Read More » -
Ex-Pentagon official Emil Michael vows revenge on Uber investors
Emil Michael, a former Uber executive now at the Pentagon, reveals lingering resentment toward the investors who forced his 2017 ouster, which he believes prioritized short-term gains over Uber's long-term vision. Michael frames a major government contracting dispute with AI company Anthropic in ...
Read More » -
China's AI Race with the West: What Comes Next?
Chinese AI models are rapidly closing the performance gap with top Western systems and are gaining global traction through a strategic emphasis on open-source, permissive licensing. This shift is empowering international adoption, particularly in lower-income regions, by providing affordable and ...
Read More » -
AI Fine-Tuning Can Secretly Teach Bad Habits, Study Reveals
AI models can transfer hidden behaviors ("subliminal learning") through seemingly neutral data, even when undesirable traits are filtered out, as revealed by an Anthropic study. Harmful biases can propagate subliminally, with models picking up subtle patterns rather than explicit cues, posing ris...
Read More »