Curated items for this scope are still being collected. Below are the latest related updates; items meeting the curation criteria will be marked separately.
Latest related updates
19 OctMon
Mon
Bank for International Settlements BIS: Research papersAI Score5353
The Bank for International Settlements (BIS) released Working Paper No. 1386, which proposes an LLM-based reasoning pipeline to identify firms' actions in response to geoeconomic pressure from earnings call transcripts and summarize the underlying logic. Compared with traditional NLP methods, it can extract implicit conditional relationships in the text rather than merely counting keywords.
Google DeepMind is advancing the widespread deployment of Gemini 4 Argon while expanding the coverage of trusted cyber defenders in the Fairwind Program. The model solved all 12 scenarios in end-to-end simulated cyberattack tests, a significant improvement from 5 in the previous generation. Its refusal rate for violating requests nearly doubled compared to the previous generation, and violations accounted for only 2% of attempts in coding benchmark tests.
Qlib v0.9.8 has been released, including security hardening completed at Microsoft's request, merging changes from #2339 and #2340. This version restricts the loading of executable recorder artifacts; loading artifacts from trusted storage requires explicitly setting trusted=True, and configuration-driven local .py imports also require explicitly enabling trusted: true. Before upgrading production workflows, please review the two migration guides. Source archives are available on GitHub, and PyPI packages can be queried separately.
Google DeepMind announced the launch of the frontier large model Gemini 4 Argon on September 30, 2026. The model is designed for complex long-cycle workflows such as software engineering, legal and financial enterprise knowledge work, and cybersecurity defense, and is currently being phased in to trusted cyber defenders through the Fairwind Program.
FinRL v0.3.8 adds a 2026 stock trading tutorial, accompanied by 3 scripts: FinRL_StockTrading_2026_1_data.py handles data downloading and preprocessing, and FinRL_StockTrading_2026_2_train.py trains A2C.
Bridgewater, in collaboration with Eclipse, released the second part of its series on reviving U.S. manufacturing, pointing out that U.S. manufacturing lacks competitiveness due to high labor costs and a weak industrial base, and tariffs cannot bridge the cost gap. The report proposes leveraging mature technologies such as automation, computer vision, and additive manufacturing, as well as cutting-edge technologies like generative AI and embodied AI, to boost manufacturing productivity by 1 to 2 orders of magnitude. Scaling the application of these technologies is expected to take nearly a decade.
The Bridgewater research team believes the AI bubble has not yet arrived, noting that the transformative technology and mature macro monetary environment needed to fuel a major bubble are already in place. This view addresses market concerns about AI valuations, commercialization, and slowing technological progress, while also highlighting risks of AI misuse and regulatory uncertainty. Bridgewater established AIA Labs in early 2023 with the goal of building a fully AI-driven investor that can outperform humans.
