Australian government bond yields hit their highest level since 2011.

According to market sources, SpaceX’s AI team will build a company from scratch next week, and the entire process will be livestreamed.

Beating AI Express, Anthropic released a new AI abuse investigation report, alleging that Moonshot AI (developer of Kimi) had directly forwarded some Kimi users’ requests to Claude, then displayed Claude’s responses to those users. Anthropic said that within a 10-day window, nearly 300,000 user requests were forwarded in this manner, with the vast majority routed to its Opus model. Moonshot used 5,380 unauthorized accounts for this scheme, and the affected users believed they were still using Kimi at the time. Anthropic also alleged that Moonshot saved some user interactions and used "cross-session replay" to extract Claude’s full reasoning process for training its own models. Between May and July, over 23 million such distillation interactions took place. Some actual user requests included internal corporate code and valid credentials, which were also forwarded to Claude. Anthropic noted it was unaware if these users had been notified that their data would be sent to a third party. These claims currently stem primarily from Anthropic’s internal investigation and have not been independently verified. In response to similar distillation allegations raised earlier by three U.S. institutions, China’s Ministry of Commerce stated the claims are "unfounded and legally baseless," adding that distillation is a common practice in the industry.

According to Bitget market data, the Nikkei 225 index fell 3.00% intraday, currently trading at 63,229.63 points. Kioxia shares dropped 6.6%, SoftBank Group declined 6.3%, and Advantest fell 6.5%.

Beating AI Express: Cognition, the parent company of coding agent Devin, has launched a new programming model, SWE-2. Built directly on Moonshot’s 2.8T-parameter Kimi K3 and further trained via reinforcement learning, SWE-2 delivers a 5-6 percentage point improvement across multiple programming benchmarks. On Cognition’s own FrontierCode 1.1 Main benchmark, SWE-2 scores 50.0%, outperforming GPT-5.6 Sol (47.5%) and Grok 4.6 (48.0%). It lags Fable 5.1 (50.9%) by just 0.9 percentage points, yet is 64% cheaper. GPT-6 Astra scores 53.3%; SWE-2 trails by 3.3 percentage points while costing roughly a quarter of Astra’s price. SWE-2 also brings major efficiency gains over its predecessor: under medium inference intensity, it cuts interaction rounds for task completion by 58% and reduces average costs by 81%. However, on the more challenging Terminal-Bench 4, SWE-2 only scores 27.3%, notably lower than Astra’s 57.9% and Fable 5.1’s 55.8%, indicating it has not yet fully matched leading models. SWE-2 is now live on Devin Desktop and CLI, and is being rolled out to Devin Web and Fusion.

Data from South Korea’s customs shows that semiconductor exports rose 270.1% year-on-year between September 1 and 10.

According to Bitget market data, South Korea’s KOSPI index dipped to 7,000 points, down 3.29% at the opening. SK Hynix and Samsung Electronics both fell by around 4%.