Latest stories

Latest stories

• 16 Oct 2026

  Taiwan

  Taiwan plans criminal penalties for unauthorised AI chip exports to China

  Taiwan plans to add advanced semiconductors and AI servers to its export control list, Economics Minister Kung Ming-hsin said in a Monday interview with Bloomberg, as reported by Taiwan News. The minister said unauthorised shipments of restricted equipment could then lead to criminal charges under the Foreign Trade Act. Kung said the changes are part of the commitments under the Taiwan-US Agreement on Reciprocal Trade and that some export controls would follow US rules, but he did not say when they would take effect. Taiwan News reported that in August prosecutors indicted nine people over an alleged scheme to export AI servers to China, using charges including breach of trust and document forgery.

  Source: Taiwan News, citing Bloomberg

• 16 Oct 2026

  International

  G7 finance leaders agree drills against AI-driven cyber attacks

  G7 finance ministers and central bank governors agreed to hold tabletop exercises simulating AI-driven cyber attacks on the financial system, Japanese Finance Minister Satsuki Katayama told reporters after their meeting on the sidelines of the IMF annual gathering in Bangkok, Reuters reported. A Japanese finance ministry official said the results will test existing protocols and be used to revise manuals, and the exercises could include private financial institutions depending on the simulation. The IMF warned this week that an AI-enabled cyber attack on a critical service of a financial institution could undermine confidence.

  Source: CNA (Reuters report)

• 16 Oct 2026

  Global

  Reflection releases open weights for its 501B-parameter Beam model

  Reflection released the weights of Beam, its first open-weight model, on Hugging Face as reflection/Beam-1.0-501B-A23B under the Apache 2.0 licence, together with a model card and a safety technical report. Beam is a sparse mixture-of-experts model with 501 billion total and 23 billion active parameters, built for coding, reasoning and agent work. Weights come in BF16, FP8 and NVFP4 formats, and the model is also offered through Reflection's API (public beta) and partners including AWS, Google Cloud, Microsoft Foundry and Databricks.

  Source: Reflection

• 16 Oct 2026

  United States

  Anthropic pitches a draft federal frontier AI safety bill

  As reported by POLITICO, Anthropic representatives have privately shown Capitol Hill lawmakers a draft bill on frontier AI safety and security. Under the draft, the government could block a model's release or use only where it poses a foreseeable, imminent risk of more than $1 billion in damage, or of killing or seriously injuring more than 50 people, arising from a biological weapon, the model evading its developer's controls, or a cyberattack. The draft would also preempt some state AI safety laws. POLITICO reported that Anthropic said the draft reflects its Advanced AI Framework, which it has promoted to federal and state lawmakers since June.

  Source: POLITICO