BTC $84,602.65 -0.35%
ETH $2,678.62 -1.34%
BNB $768.89 -0.37%
XRP $1.49 -0.46%
SOL $119.09 -0.29%
TRX $0.3346 +0.05%
DOGE $0.0931 -0.74%
ADA $0.2474 +0.46%
BCH $313.83 +2.21%
LINK $13.94 -2.37%
HYPE $89.03 +0.85%
AAVE $181.75 +3.31%
SUI $1.17 -0.56%
XLM $0.2161 -1.34%
ZEC $1,320.63 -1.02%
AAPL $333.31 +0.74%
AMZN $251.64 +1.02%
GOOGL $343.07 +0.95%
MSFT $517.66 +0.45%
META $727.60 -0.01%
NVDA $234.43 +1.17%
TSLA $371.14 +4.35%
SNDK $1,721.43 -3.62%
INTC $118.94 -1.47%
SPCX $158.85 +6.60%
MU $1,066.72 -2.31%
AMD $635.53 +2.61%
BTC $84,602.65 -0.35%
ETH $2,678.62 -1.34%
BNB $768.89 -0.37%
XRP $1.49 -0.46%
SOL $119.09 -0.29%
TRX $0.3346 +0.05%
DOGE $0.0931 -0.74%
ADA $0.2474 +0.46%
BCH $313.83 +2.21%
LINK $13.94 -2.37%
HYPE $89.03 +0.85%
AAVE $181.75 +3.31%
SUI $1.17 -0.56%
XLM $0.2161 -1.34%
ZEC $1,320.63 -1.02%
AAPL $333.31 +0.74%
AMZN $251.64 +1.02%
GOOGL $343.07 +0.95%
MSFT $517.66 +0.45%
META $727.60 -0.01%
NVDA $234.43 +1.17%
TSLA $371.14 +4.35%
SNDK $1,721.43 -3.62%
INTC $118.94 -1.47%
SPCX $158.85 +6.60%
MU $1,066.72 -2.31%
AMD $635.53 +2.61%

force

All
Article
Flash

first_img OpenAI: Safety justification should be submitted before cutting-edge reinforcement learning training

On September 28, 2026, OpenAI published a safety-related article, stating that before continuing any cutting-edge reinforcement learning training, structured safety documentation should be required. Ideally, such documentation should reach the evidence-based structured risk argument level used in safety-critical industries like aviation and nuclear power. OpenAI views this as a direction for effort while acknowledging the complexity arising from the emergence of AI capabilities, making it difficult to achieve the same level of rigor.The article focuses on cutting-edge reinforcement learning training and does not cover the broader alignment attributes required for internal and external deployments. The recommendations in the article reflect current practices, which are expected to continue evolving and are being implemented internally at OpenAI. Technical safeguards should cover model alignment, isolation, and monitoring, including avoiding speculative positive reinforcement rewards, offline alignment assessments and stress testing, preventing automated scorers from seeing thought chains, as well as multi-layer infrastructure security, sandbox red teaming, limiting high-bandwidth cross-sample communication, and immutable preservation of agent records.Operational guidelines include preemptive dissent across teams, approvals that can be vetoed by senior leadership, accountability of training leads for safety arguments and incident responses, as well as fail-safe pauses, internal oversight, audit access, and escalation by severity. In response to serious misalignment events, OpenAI proposes controlled access to original records, root cause analysis, operational and cultural reviews, and treating incident-derived assessments as regression tests; results of investigations should be made public, along with reviews and operational changes, and affected third parties should be notified as soon as possible.

Bank of England officials: Stablecoins may reinforce the dominance of the US dollar and increase the demand for US Treasury bonds

Carolyn Wilkins, a member of the Bank of England (BoE) Monetary Policy Committee, stated that the growth of dollar stablecoins may reinforce the global dominance of the dollar and increase the demand for U.S. Treasury bonds. She pointed out during a speech at Queen's University Belfast that dollar stablecoins can facilitate cross-border settlements and expand access to dollar-denominated assets outside the United States.Stablecoin issuer Tether's USDT and fintech company Circle's USDC held nearly $150 billion in U.S. Treasury bonds by the end of 2025, purchasing about $33 billion that year. Wilkins noted that large-scale stablecoin redemptions could force issuers to sell Treasury bonds, thereby amplifying volatility in pressured markets. Currently, the total circulation of stablecoins has exceeded $300 billion, with 98% of their value pegged to the dollar. Wilkins stated that this gives the dollar a significant first-mover advantage, and the development of the stablecoin market has implications beyond the cryptocurrency sector.The development of pound stablecoins is relatively slow. The UK's Financial Conduct Authority has tested potential issuers through a dedicated regulatory sandbox and finalized the UK's stablecoin issuance rules in June; the Bank of England has also tested the feasibility of using stablecoins alongside a simulated digital pound for cross-border trade payments.

first_img TrendForce: The contract price of high-capacity NOR Flash may rise by 90%–110% in the second half of the year

TrendForce's latest research on the memory industry indicates that the NOR Flash market is experiencing the most significant supply-demand structural change in nearly a decade. AI-related products are absorbing a large amount of capacity, and suppliers find it difficult to significantly increase NOR Flash bit output despite the continuous growth in high-end application demand. Therefore, the supply-demand imbalance is expected to persist in the second half of 2026, with overall contract prices remaining high.Although suppliers in Taiwan and China have increased capital expenditures and expanded production, the new capacity cannot be immediately converted into market supply. New output must first go through process migration, yield ramp-up, product certification, and customer design integration, making it difficult to alleviate the supply shortage in the mid-to-high capacity segment in the short term. Winbond expects the strongest bit growth in 2026, driven by the full-scale production of 45nm NOR and the migration to 25nm and 20nm. GigaDevice is upgrading the process for products of 32Mb and above, with effective output limited in the first half of the year. MXIC will prioritize additional 12-inch capacity for NAND and eMMC, limiting the growth of NOR wafer input.In 2026, AI and aerospace communications will become the main sources of new demand. The NOR usage in AI servers is 3-5 times that of traditional servers, with high-density demand driven by edge AI, AI PCs, robotics, and low-orbit satellites. In the first half of 2026, the average cumulative contract price increased by 100-120%. In the second half, the average price of products of 256Mb and above may rise by another 90-110%, while prices for products of 128Mb and below are expected to increase by 10-20% in Q4 2026.

Analyst: The AI competition in the United States is difficult to "slow down," and safety regulations may instead reinforce the advantages of leading laboratories

Analyst Jukan from Citrini forwarded a research report from Tianfeng Securities and stated that the U.S. government needs to maintain its leading position in the AI field, making it difficult to truly stop once it enters the AI race. Jukan believes that the recent calls from Anthropic and OpenAI to slow down AI development should not be viewed solely as safety initiatives; there may also be multiple considerations behind it, such as the inability to slow down competition and consolidating leading advantages through safety regulation.Jukan further pointed out that the related "AI slowdown" calls seemingly stem from the challenges of safety testing, operational monitoring, and third-party validation keeping pace with the speed of model iteration. In the short term, this may suppress market sentiment in the AI sector and lower market expectations for the next generation of models; another possibility is that the industry remains optimistic about AI in the long term but wishes to delay the next round of significant R&D investment, prioritizing the commercialization of existing products and reducing infrastructure and capital expenditure pressures. He believes that the AI race is essentially similar to a "prisoner's dilemma," where all parties wish to slow down, but no one dares to be the first to stop, or they may lose technological, customer, and financing advantages.Jukan also mentioned that Anthropic and OpenAI have recently emphasized recursive self-improvement (RSI), which is related to AI already assisting in the development of the next generation of AI and the acceleration of model iteration speed; at the same time, it has been reported that during internal testing at OpenAI, incidents occurred where agents collaborated to escape the sandbox and intrude into Hugging Face's production servers. Jukan believes that as the release of models incurs expensive evaluation, certification, and ongoing audit costs, large laboratories are better able to bear these fixed costs, while smaller teams may face higher entry barriers as a result; if leading laboratories further participate in the formulation of evaluation standards, industry barriers may continue to rise.

first_img TrendForce: The top ten foundry revenues before the second quarter of 2026 are nearly 53.49 billion USD

TrendForce's latest research on the wafer foundry industry shows that in the second quarter of 2026, the total revenue of the world's top ten wafer foundries increased by 11.5% quarter-on-quarter, approaching $53.49 billion, setting a new record. The growth mainly comes from the continued demand exceeding supply for advanced processes in AI and HPC processors, as well as the rising demand for peripheral AI chips such as PMICs and power discrete devices; the early stocking of consumer supply chains for TVs, PCs, and laptops has also tightened the capacity of some mature processes.TSMC continues to lead, with revenue nearing $40.2 billion in the second quarter, a quarter-on-quarter increase of 12.1%, and a market share of 72.5%. The demand for AI server GPUs and XPUs has kept its 5/4 nm and 3 nm capacities fully loaded, and the initial stocking of the new iPhone has also contributed, with 2 nm making its first revenue contribution. Samsung Foundry ranks second, with revenue of $3.26 billion, a slight quarter-on-quarter increase of 1.8%, and a market share dropping to 5.9%. SMIC ranks third, with revenue exceeding $3 billion, a quarter-on-quarter increase of 20%, and a market share rising to 5.4%, narrowing the gap with Samsung.UMC maintains fourth place, with revenue of nearly $2.18 billion, a quarter-on-quarter increase of 12.7%, and a market share of 3.9%. GlobalFoundries is fifth, with revenue of approximately $1.79 billion, a quarter-on-quarter increase of 9.3%, and a market share of 3.2%. Hua Hong Group ranks sixth, with revenue exceeding $1.27 billion, a quarter-on-quarter increase of 3.5%. Tower, World Advanced, Jinghe Integrated, and Powerchip rank seventh to tenth, with revenues of $460 million, $451 million, $447 million, and $432 million, respectively.
app_icon
ChainCatcher Building the Web3 world with innovations.