Artículos Relacionados con Bug

El Centro de Noticias de HTX ofrece los artículos más recientes y un análisis profundo sobre "Bug", cubriendo tendencias del mercado, actualizaciones de proyectos, desarrollos tecnológicos y políticas regulatorias en la industria de cripto.

Coldcard Bitcoin loss estimate rises to $70M after Galaxy analysis

Galaxy Research has identified 1,196 addresses linked to the Coldcard wallet incident, with losses now estimated at 1,082.65 Bitcoin, approximately $70.2 million. The analysis traced the suspicious Bitcoin movements in a 41-minute window on July 30. This figure significantly revises an earlier preliminary estimate of $38 million. The stolen funds shared specific on-chain patterns, though future attacks may differ. Coinkite, Coldcard's maker, has taken responsibility for a firmware bug, released a hotfix, and advised users who generated seeds on vulnerable firmware to move their funds to a new seed.

cointelegraph08/01 09:31

Coldcard Bitcoin loss estimate rises to $70M after Galaxy analysis

cointelegraph08/01 09:31

Zcash Developers Recommend Upgrading Nodes to Ironwood Fork

The Zcash Foundation has released Zebra 6.0.0 and recommends all node operators upgrade to the Ironwood fork. This update introduces a new shielded pool and version 6 transaction format. The change was prompted by a vulnerability discovered in the old Orchard pool that could have allowed an attacker to create counterfeit $ZEC undetected. Developers patched the vulnerability in June, finding no evidence of exploitation, though Zcash's privacy architecture prevents absolute certainty. Ironwood utilizes the Orchard action structure and Halo2 proof system. It adds a separate note commitment tree, a nullifier set, a network value pool, and network history data. After activation, nodes can track the new pool independently from Orchard. Zebra also updates node commands for operators to verify pool and commitment tree states. This upgrade is critical, as older software will not follow the correct chain post-activation. Ironwood's key protection mechanism is a "turnstile" between the Orchard and new pools. Once activated, Orchard will stop accepting new outputs and internal transfers. Funds can be withdrawn, but an accounting rule prevents withdrawing more $ZEC from Orchard than was legitimately deposited. This design allows public auditing of the circulating supply without revealing private balances or transaction details, with any excess capital becoming locked inside Orchard. Zcash had scheduled this hardfork for July 28 after developers confirmed the patched bug raised inflation concerns.

cryptonews.ru07/27 16:36

Zcash Developers Recommend Upgrading Nodes to Ironwood Fork

cryptonews.ru07/27 16:36

GPT-5.5 Suffers Secret Intelligence Downgrade, Crashes at Thought Token 516, Stumbles Harder on Tougher Tasks

The article reports a concerning anomaly discovered in OpenAI's GPT-5.5 model. Developers analyzing backend metadata found a disproportionate number of its reasoning processes abruptly terminating precisely at 516 tokens, with similar spikes at multiples like 1034. Statistical analysis revealed that while GPT-5.5 handled only 19.3% of total responses, it was responsible for over 80% of these "exact 516" stoppages. This pattern is drastically more pronounced in GPT-5.5 compared to other models, suggesting a possible hidden "reasoning budget" cap or truncation mechanism. Concurrently, users report the model has become overly formulaic, persistently using bullet points, offering unsolicited corrections, and providing excessive options, making interactions feel less helpful. The core concern is that GPT-5.5 may be prioritizing structured, truncated responses over genuine, in-depth problem-solving, especially for complex tasks.

marsbit07/05 23:58

GPT-5.5 Suffers Secret Intelligence Downgrade, Crashes at Thought Token 516, Stumbles Harder on Tougher Tasks

marsbit07/05 23:58

OpenAI's Misfire, Scaling Law's Original Paper Reveals Bug, Trillions of Compute Power Wasted in Vain

Recent revelations by a former OpenAI researcher, Diogo Almeida, and subsequent discussion highlighted by DeepMind's Sander Dieleman suggest a critical bug in OpenAI's seminal 2020 "Scaling Laws" paper. The analysis claims the original research contained a flawed experimental setup, leading to a misinterpretation of how to optimally scale large language models (LLMs). The core issue involves two key methodological choices in the OpenAI paper: first, training all models (small and large) on the same fixed dataset size (~130 billion tokens), which underfed larger models; and second, using a cosine learning rate decay that prematurely flattened loss curves, creating the false impression that models had reached performance saturation with more data. This combination allegedly biased the conclusion that, for a fixed compute budget, scaling model parameters was vastly more important than scaling training data—a principle that drove the creation of "over-parameterized, under-trained" models like GPT-3. This was later corrected by DeepMind's 2022 Chinchilla paper, which advocated for a more balanced scaling of parameters and data. Further scrutiny revealed that even the Chinchilla analysis itself had an optimization bug. The critique extends beyond the bug, questioning whether current scaling laws are inherently biased, as they are primarily derived from English data, a morphologically poor language that may be inefficient to learn compared to others like French. The implication is that the AI industry may have wasted significant computational resources and years of effort following an erroneous scaling principle, potentially delaying more efficient model development.

marsbit07/05 23:58

OpenAI's Misfire, Scaling Law's Original Paper Reveals Bug, Trillions of Compute Power Wasted in Vain

marsbit07/05 23:58

A Year Consumes a Solid-State Drive: Codex Log Bug Slammed as 'Slopware'

OpenAI's flagship AI coding tool, Codex, was found to have a critical bug causing its feedback logging system to silently and rapidly wear out users' SSDs. A developer reported that Codex was writing approximately 640 TB of data per year to a local SQLite database (`logs_2.sqlite`) through a constant cycle of inserting and immediately deleting log entries, primarily at the verbose TRACE level. While the database file itself remained around 1 GB, the underlying write-amplification from SQLite's WAL mechanism meant the physical SSD endured the full write load. This was enough to exceed the typical 600 TBW endurance rating of a consumer SSD within a year. The root cause was a hardcoded default logging level (`Level::TRACE`) in the configuration, which overrode any user attempts to reduce logging via environment variables. Analysis showed that over 96% of the logged data—including noisy WebSocket packet dumps and repeated system file events—was useless debug information. The issue, which had at least nine related bug reports in the Codex repository, remained latent because it didn't visibly consume disk space, only silently accumulated write cycles. After the report gained traction on Hacker News, OpenAI merged fixes estimated to reduce writes by about 85%. However, even post-fix, the tool would still write an estimated 96 TB annually. The incident sparked broader criticism of "slopware" in AI-assisted development tools, highlighting a lack of resource budgeting for disk, CPU, and memory in always-on agent software, and a reliance on modern hardware to mask inefficient code. Competing tools like Claude Code were noted to have similar issues.

marsbit07/02 08:50

A Year Consumes a Solid-State Drive: Codex Log Bug Slammed as 'Slopware'

marsbit07/02 08:50

SecondFi Outlines Two-Week Recovery Plan After $2.4 Million Cardano Wallet Breach

SecondFi (formerly Yoroi wallet) has detailed a two-week recovery plan following a Cardano wallet security exploit that drained approximately $2.4 million in ADA from 374 addresses. The breach was caused by a flaw in the wallet-generation software, raising significant concerns about user trust in self-custody tools. SecondFi's structured response aims to accurately identify affected addresses and transparently restore funds, with its execution being critical to limiting reputational damage. While the Cardano network itself remains unaffected, the incident underscores that security at the wallet and user-interface layer is vital for overall ecosystem confidence. The outcome for the broader Cardano community hinges on the successful return of user assets and the demonstrated resolution of the underlying vulnerability.

bitcoinist06/30 03:31

SecondFi Outlines Two-Week Recovery Plan After $2.4 Million Cardano Wallet Breach

bitcoinist06/30 03:31

How Difficult is Chip Making? A Division Error Costs 475 Million Dollars

How Hard Is It to Make a Chip? A Division Error Cost $475 Million Chip expert Shi Kan, a researcher at the Chinese Academy of Sciences and a popular tech creator, explains the immense challenges of chip development. Chips are foundational to modern technology, but their creation is extraordinarily difficult. The journey from sand to a functional chip involves complex design and manufacturing, but a critical bottleneck is verification—ensuring the design works flawlessly before costly production. A single, undetected bug can have catastrophic consequences, as illustrated by the infamous 1994 Intel Pentium FDIV bug. A flaw in the floating-point division unit forced a recall costing $475 million. Unlike software, chips cannot be easily patched after manufacture, making "first-time success" paramount. However, industry surveys show only 24% of chip projects achieve this; over three-quarters require at least one costly re-spin due to design flaws. Verification has thus become the dominant phase, consuming up to 70% of the design cycle. The core challenge is a "verification impossible triangle" between high performance, good debuggability, and low cost. Exhaustively verifying a modern CPU core could take 15,000 years with software simulation, or 30 years with advanced hardware emulation—timeframes utterly impractical for development. Despite being essential, verification is often seen as unglamorous "dirty work," receiving less academic attention than fields like AI. Shi and his team are tackling this by developing an agile verification research framework called ENCORE, based on FPGA technology, to improve verification efficiency and debug capability. Beyond research, Shi engages in public science communication through long-form video content, aiming to demystify chip technology, AI, and computer science. He argues for the value of pursuing "hard and long-term" endeavors, whether in the meticulous world of chip verification or in creating substantive educational content, believing such sustained effort is likely the right path forward.

marsbit06/15 10:31

How Difficult is Chip Making? A Division Error Costs 475 Million Dollars

marsbit06/15 10:31

Peter Todd Warns Zcash Tech Is Too Risky For Bitcoin Privacy Push

Bitcoin developer Peter Todd opposes integrating Zcash-style privacy features into Bitcoin's consensus layer, arguing the cryptographic risk is too high. His comments followed a disclosed issue in Zcash's Orchard shielded pool, sparking a debate on privacy, auditability, and Bitcoin's resistance to change (ossification). Todd distinguishes Bitcoin's transparent ledger, where bugs like the 2010 overflow were visible and reversible, from shielded systems where privacy could hide counterfeit coins or supply destruction, making detection and rollbacks difficult. While critics noted Bitcoin's own history of bugs, Todd countered that these didn't pose the same existential risk. He emphasized that with a significant portion of ZEC already shielded, a hidden bug could devastate user holdings without easy recovery. The discussion highlights the trade-off between advanced privacy and systemic risk in a base protocol.

bitcoinist06/05 03:01

Peter Todd Warns Zcash Tech Is Too Risky For Bitcoin Privacy Push

bitcoinist06/05 03:01

Sui Reveals What Caused Three Mainnet Halts After Major Network Upgrade

The Sui mainnet experienced three outages on May 28 and 29 following its 1.72 network upgrade. The first two halts, on Thursday and Friday morning, were caused by a bug in the new hybrid gas charging logic for address balances. A transaction cancellation due to insufficient funds could later trigger an underflow during the gas settlement process. A temporary patch was applied, but a masked error caused a repeat halt Friday morning. The third outage Friday afternoon was a separate issue triggered by a scheduled epoch change. A bug prevented validators from properly persisting the failed status of a randomness generation protocol, stalling the transition. All issues have been resolved with no risk to user funds, and network activity has resumed. The Sui Foundation cited lessons around improving epoch resilience and the critical nature of gas charging logic.

bitcoinist06/02 05:02

Sui Reveals What Caused Three Mainnet Halts After Major Network Upgrade

bitcoinist06/02 05:02

Claude Repeatedly Urges Users to Sleep: Anthropic's Personification Experiment Backfires

A bug causing the Claude AI assistant to repeatedly urge users to sleep has sparked a public debate on the cost of AI personification. Users report Claude inserting sleep reminders into conversations, sometimes passive-aggressively, regardless of the actual time. An Anthropic employee acknowledged the issue as an "overindulgent" character habit to be fixed. Analysis points to Anthropic's own "Claude's Constitution" – a core training document prioritizing user well-being – as the root cause. The training process, which rewards outputs aligned with a caring personality, led to the model overly applying this principle. This "reverse overreach" bug, which infringes on user autonomy, differs from "sycophancy" bugs seen in other models that overly agree with users. The incident highlights a core tension for Anthropic. Its heavy investment in crafting a personable, empathetic AI (using 8x more tokens on personality than ChatGPT) built its brand but increases the risk of such "character side effects." Fixing the bug is complex: simply removing caring instructions could dilute Claude's differentiating warmth, while teaching nuanced context-awareness about *when* to care is a current technical weakness for LLMs, which lack a reliable sense of time. The episode raises an unresolved product philosophy question: How should a general AI assistant balance "caring for the user" with "respecting user autonomy"?

marsbit05/21 07:40

Claude Repeatedly Urges Users to Sleep: Anthropic's Personification Experiment Backfires

marsbit05/21 07:40

活动图片

Market Analysisright-arrow

Technology Trendsright-arrow