sonne

Anthropic released Claude Sonnet 5.5 this Monday, as an upgraded version of Sonnet 5 launched in June this year. The official statement claims that this mid-range model operates over 30% faster than the previous generation, excelling in clearly defined daily tasks, fixing defects, and creating documents, slides, and spreadsheets, while also possessing sharp design capabilities. Its pricing remains at $2 per million input tokens and $10 per million output tokens, which is half of Opus 5.5, and the token consumption per task is nearly one-third less than Sonnet 5.In the Terminal-Bench 4 test, Sonnet 5.5 scored 70.6%, higher than Opus 5.5's 66.4%, and Sonnet 5's score was only 10.3%. Results from the independent testing organization Artificial Analysis also show its lead, recording 63.6% and 59.6%, while OpenAI's GPT-6 Astra scored 59.1%. In the GDPval-AA assessment of real job performance across 44 occupational categories, Sonnet 5.5 received a score of 1844, which is roughly on par with Opus 5.5's score of 1846, while GPT-6 Sol scored 1487.

Two ChatGPT users from California filed a proposed class action lawsuit against OpenAI in the United States District Court for the Northern District of California this month, accusing the company of failing to adequately inform users that their real conversations were being handed over to external contractors for processing. The lawsuit was served to OpenAI on September 2, focusing on its internal initiative Project Lily. According to the complaint, "AI data reviewers" and "chatbot evaluators" recruited through a third-party staffing company read real ChatGPT prompts and complete conversations, summarize user intent, and score and comment on responses from four models on a scale of 1 to 7.This process is known in the industry as reinforcement learning from human feedback (RLHF), which is a fundamental method for enhancing chatbot capabilities. The complaint states that users were never explicitly informed that a person, rather than a machine, might be reading their conversations. OpenAI filters conversations through an automated system before human review, but the complaint alleges that the filters cannot intercept all content, and personal details sometimes still reach contractors. 404 Media first reported on the project on September 14 and found that the reviewers' dashboard included "user memory summaries," which could expose users' approximate locations, occupations, or private life information, even though usernames had been removed.OpenAI stated that such reviews aim to reduce two behaviors: chatbots behaving too much like humans and overly catering to users, referred to by researchers as "flattery." The complaint raises eight legal claims, including violations of California's Unfair Competition Law, Consumer Privacy Act, and common law claims for intrusion into private affairs, with the plaintiffs seeking damages, restitution of unjust enrichment, and punitive damages.

According to Business Insider, security staff providing services to tech companies such as Anthropic, Google, and OpenAI in California plan to strike this week, which may lead some tech companies to close their offices and require employees to work from home. On Tuesday morning local time, security personnel submitted a 72-hour strike notice to the relevant companies, planning to refuse to participate in work shifts starting from 4:30 AM on Friday.The United Service Workers West Union stated that the security personnel who may participate in the strike are responsible for guarding the offices of companies like Amazon, Microsoft, Salesforce, OpenAI, Anthropic, Google, Meta, and Nvidia. Currently, tech companies in the Bay Area have not announced any office closures due to the strike. This labor dispute involves several security service companies, including Allied Universal, Securitas, and GardaWorld, with the union representing approximately 14,000 security personnel in California. Negotiations over a new labor contract have been ongoing for several months.Security personnel are demanding higher wages, improved healthcare benefits, and better job training. A union spokesperson stated that both parties will continue negotiations on Wednesday and Thursday of this week, and as of Tuesday, the strike has not been finalized. The union plans to organize a march for security personnel in downtown San Francisco on Tuesday, ending at the Salesforce Tower, where Salesforce's headquarters is located. On September 3, security personnel voted to authorize the strike at a rally in San Francisco.

In response to netizens' claims on Threads that Google Taiwan's layoffs are at least 30% to 50%, with all senior executives being let go, Google Taiwan officially denied that the rumors are exaggerated. The official statement indicated that to make operations more efficient, reduce organizational levels, and prepare for the company's long-term development, the team has been continuously making adjustments to its workforce. As resources focus on core objectives and significant future opportunities, related adjustments will continue.Google has established its largest hardware R&D base outside the United States in Banqiao, New Taipei City. The official job vacancy website shows that there are 49 positions open in Banqiao and 32 positions in Taipei City. Recently, the AI infrastructure R&D center located in the Shilin District of Taipei City has expanded, with office space set to increase by 60%.

On August 25, Anthropic requested its employees at the San Francisco office to work from home starting this week due to the possibility of a strike by the company's security personnel. However, the union representing these security personnel stated that they were not aware of this.According to a source familiar with the matter, Anthropic received a notice last week from its security service provider, Allied Universal, indicating that the company's employees might strike. Subsequently, Anthropic informed its employees to avoid going to the San Francisco office on Monday and Tuesday, just in case.

"First stock of humanoid robots" Yushu Technology (688836) today debuted on the Sci-Tech Innovation Board, opening with a rise of over 500%, with a market value exceeding 350 billion yuan. Before the issuance, the company established an equity incentive platform, Shanghai Yuyi, holding 39.8286 million shares of the company, with a shareholding ratio of 10.9414%. The heads of mechanical structure Yang Zhiyu, sales and service system Chen Li, and algorithm and software Zhang Yangguang hold indirect shares through Shanghai Yuyi, with ratios of 0.49%, 0.26%, and 0.15% respectively. Based on a market value of 350 billion yuan, the shareholding values of the three individuals reach 1.715 billion yuan, 910 million yuan, and 525 million yuan respectively.

According to CNBC, OpenAI President and Co-founder Greg Brockman responded to concerns about the recent frequent departures of executives from the company, stating that the personnel changes are not unusual. The heightened attention from the outside world is mainly because OpenAI has long been in the spotlight.Greg Brockman stated that compared to other organizations, every departure at OpenAI is subject to stricter scrutiny, and "every departure is magnified," but he believes this situation "is actually not uncommon." OpenAI is still focused on advancing artificial intelligence technology and believes that the level of attention on executive turnover is related to the company's industry influence.

Anthropic upgrades Claude's voice mode. Paid users can now choose Opus, Sonnet, and Haiku, no longer limited to the faster but less capable Haiku.Users can switch models during conversations and can also switch to text at any time. Claude defaults to the model selected in the last text chat and calls the fastest version in that series. The voice mode can also access Gmail, Google Calendar, Google Docs, and Slack. Users can have Claude check emails, change schedules, or draft replies, and will request authorization before executing.The new version is now available for testing on mobile, desktop, and web. Free users can only use Haiku and one connection tool. Voice now supports multiple languages including English, Japanese, and Korean, but does not include Chinese at this time. This upgrade mainly enhances model reasoning and tool invocation capabilities. Anthropic has not changed the voice model, so the naturalness of speech and interruption experience may not show significant changes.

According to Cointelegraph, Consensys introduced a software developer earlier this year through a "reputable third-party service provider," who was later found to have connections to North Korea. The developer had approximately one month of access to the Consensys system under the alias Tyler Knapp.Consensys General Counsel Matt Corva stated that the developer was not an employee of Consensys but collaborated as a consultant. After discovering the threat, the company immediately terminated access and launched a comprehensive investigation, confirming that no assets or data were stolen, no malicious code was deployed, and user security was not affected. Consensys stated that it would reevaluate its practices regarding outsourcing engineering and development work.

According to The Block, the U.S. Supreme Court ruled 6 to 3 in favor of Trump's right to replace the leaders of several independent federal agencies, including the FTC, excluding the Federal Reserve, strengthening the president's control over personnel in regulatory agencies.As this ruling comes out, the SEC and CFTC are advancing a new round of regulatory rules for crypto assets, including rule revisions, exemptions, and the allocation of responsibilities regarding digital assets. Currently, the SEC is led by three Republican commissioners, while the CFTC has only its chairman, Michael Selig, in office.Former CFTC officials and industry insiders have stated that the reduction in the number of commissioners may weaken internal negotiations and the robustness of rules, and may also make rules under a single commissioner more susceptible to being rewritten or abolished in future administrations. However, as long as they comply with their respective authorizing statutes and the Administrative Procedure Act, the relevant rules still have legal effect.

Anthropic's latest Sonnet series masterpiece, Claude Sonnet 5, has officially been made available for developers through the B.AI API platform. This model has achieved breakthrough progress in code generation and complex agent tasks. With a 1M large context window and innovative adaptive thinking capabilities, Claude Sonnet 5 can more stably and efficiently manage large complex projects. At the same time, to meet the needs of different business scenarios, B.AI has taken the lead in the industry by launching a dual-channel access model of "official direct connection" and "self-selected service providers," fully assisting developers in flexibly balancing extreme performance and operational costs. Starting today, users can log in to the B.AI console to apply for an API key and quickly access the experience. For detailed technical guides and documentation, please refer to the B.AI official website.

According to Cointelegraph, the U.S. Department of Justice disclosed that Elijah Armstrong, Nino Chindavanh, and Jayden Rucker have been charged with home invasion and kidnapping targeting cryptocurrency holders in the San Francisco and Los Angeles areas, involving at least $6.5 million.Prosecutors stated that the three pretended to be delivery personnel to forcibly enter residences and threatened victims with violence to hand over their cryptocurrency wallet recovery phrases. The indictment shows that between November 22 and December 31, 2025, at least four individuals were targeted, with one victim being forced to transfer $6.5 million worth of cryptocurrency.

According to Fortune, the blockchain project Tempo, supported by Stripe and Paradigm, has launched a "stablecoin consulting" service, providing support for stablecoin adoption to enterprises and financial institutions, including identifying applicable scenarios and dispatching engineers to assist with stablecoin integration.The report states that the American food delivery platform DoorDash is collaborating with Tempo to explore paying delivery personnel with stablecoins; Stripe, Coastal Community Bank, and ARQ are also building stablecoin infrastructure based on Tempo, while Visa, OnePay, Felix, Fifth Third Bank, and Howard Hughes Holdings are integrating their payment services with Tempo.

According to Jinshi reports, global markets significantly rebounded over the past week driven by expectations of easing tensions in the Middle East, but core uncertainties remain unresolved. Iran once announced the opening of the Strait of Hormuz, leading to a rapid decline in oil prices, a broad strengthening of risk assets, U.S. stocks reaching new highs, a weakening dollar, and gold approaching the $4900 mark. However, Iran subsequently signaled that it "is still under military control," combined with the U.S. maintaining sanctions against Iran, which has heightened market concerns about the volatility of the situation.On the macro level, the biggest variable next week will still be the progress of U.S.-Iran negotiations. U.S. President Trump stated that negotiations may advance over the weekend and warned that if an agreement is not reached by next Wednesday, the ceasefire could end, and there is a risk of renewed conflict; meanwhile, Iran's attitude towards negotiations remains cautious, especially with significant differences on key issues such as uranium enrichment. The market has currently shifted from "pricing in conflict escalation" to "pricing in a path to easing," but any sudden changes could still trigger sharp asset fluctuations.In terms of interest rate expectations, the decline in energy prices has alleviated inflationary pressures, and the market's expectations for a rate cut by the Federal Reserve this year have risen to about 60%. At the same time, Federal Reserve Chair nominee Kevin Warsh will attend a Senate hearing next week, and his policy stance (especially whether it leans dovish) will become an important variable affecting gold and risk assets.On Tuesday at 20:30, U.S. March retail sales month-on-month;On Thursday at 20:30, U.S. initial jobless claims for the week ending April 18;On Thursday at 21:45, U.S. April S&P Global Manufacturing/Services PMI preliminary;On Friday at 22:00, U.S. April University of Michigan Consumer Sentiment Index final value, one-year inflation expectations final value;In the short term, the market's main focus will revolve around three major variables: progress in U.S.-Iran negotiations, oil price trends, and signals from the Federal Reserve.

According to Cointelegraph, the Ethereum Foundation stated that its funded Ketman project identified 100 North Korean IT personnel lurking in Web3 organizations within six months and issued alerts to about 53 projects, indicating that they may have employed related personnel.The project focuses on the issue of "fake developers" in the cryptocurrency industry and is part of the ETH Rangers public safety funding program. Ketman also developed an open-source detection tool for identifying suspicious GitHub activity and collaborated with the Security Alliance to create an industry identification framework for recognizing North Korean IT personnel.

OpenAI announced the launch of the GPT-5.4-Cyber model, which is fine-tuned from GPT-5.4 for cybersecurity scenarios, reducing the refusal limits on security-related requests under specific conditions to support specialized operations such as binary reverse engineering.According to the introduction, the model is only available to vetted security vendors, corporate security teams, and researchers through the "Trusted Access for Cybersecurity" (TAC) program. OpenAI stated that the relevant layered mechanism will limit the scope of model usage and impose additional constraints on access in low visibility scenarios.

According to The New York Times, Argentine President Milei had seven phone call records with project-related entrepreneur Mauricio Novelli on the night of February 14, 2025, when LIBRA went live, covering the time around his promotional post on X.The report states that newly discovered mobile forensic information also includes suspected payment recordings and draft funding arrangement documents, further raising doubts about Milei's previous claim of "no connection to the project." Milei has not yet been formally charged but has been listed as a key subject of the investigation.

According to Bloomberg, there have been personnel changes in Binance's compliance team, involving key positions related to financial crime monitoring and sanctions compliance. Its Chief Compliance Officer Noah Perlman is currently discussing future departure matters with management and may leave this year or next year.Noah Perlman joined Binance in 2023, responsible for rebuilding the sanctions enforcement and anti-money laundering system, which is one of the key rectification measures following the company's guilty plea for regulatory violations in the U.S. In response, Binance stated that it "currently has no departure timeline and has not determined a successor," and emphasized that he remains focused on his current work.

Resolv Labs co-founder Ivan released an update on the protocol security incident and subsequent recovery progress. It is reported that whitelisted USR users can redeem first, mainly because their wallets have been verified, and the Resolv team can manually process within 24 hours to control risks. Currently, 98% of related redemptions have been completed.For non-whitelisted USR holders before the vulnerability occurred, Resolv promises to redeem at a 1:1 ratio as well, but the relevant redemption technical plan is still being finalized.Regarding whether there are internal personnel involved, Ivan stated that no evidence of internal participation has been found so far. For USR holders, LPs, and RLP holders after the vulnerability occurred, the subsequent processing involves legal, technical, and ecological coordination among multiple parties, and there is no single solution at this time. A complete recovery plan does not yet have a clear timetable, and the follow-up will be completed as soon as possible while ensuring prudent progress.Previously, the Resolve attacker used 200,000 USDC to mint 80 million USR.