Search “Anthropic”

1Password launches password login feature for AI agents

On October 22, 2026, 1Password launched 1Password for Agents, allowing AI agents to complete logins without viewing passwords. According to 9to5Google, when an agent reaches a login form, it authenticates to 1Password using a digital key; after the user confirms the required credentials and their purpose, 1Password fills them directly into the form. The credentials are not handed to the agent and are removed once login succeeds, so the agent cannot read or reuse them. The integration currently supports only Meta's Muse and Grok Bot, with OpenAI Dots and Anthropic Claude Duet to follow. Another report says the first integrations are Meta's Muse and SpaceXSI's Grok Bot, to be deployed within the next few days; partners including Anthropic's Claude Duet, OpenAI's ChatGPT Work and dots, and Perplexity have not yet given a timeline. The service is included in personal and family subscriptions at no extra charge and currently supports only accounts hosted on 1Password.com; 1Password.eu and 1Password.ca accounts are not yet supported.

Artificial Analysis publishes AA-PersonalityBench results

On October 22, 2026, Artificial Analysis published the AA-PersonalityBench evaluation results. The results show that Anthropic models lead overall in the two categories of information density and literary quality. The highest information density score is Claude Fable 5.1 (100), followed by Claude Opus 5.5 (97), Claude Sonnet 5.5 (95) and Claude Haiku 5.5 (92); the best literary quality is Claude Haiku 5.5 (100), followed by Claude Opus 5.5 (97) and Claude Fable 5.1 (95). Currently these results are published by Artificial Analysis, and there are no subsequent developments.

ARC Prize announces verified results for Claude Sonnet 5.5 on ARC-AGI

ARC Prize announced the verified results of Anthropic's Claude Sonnet 5.5 on the ARC-AGI generations of benchmarks. For ARC-AGI-3, under the provider adapter harness the score was 72.4% at a cost of $8.2K, and under the standard harness it was 24.3% at a cost of $7.4K; ARC-AGI-2 was 83.9% at $0.55 per task; ARC-AGI-1 was 96.5% at $0.14 per task. ARC Prize said Sonnet 5.5's 72.4% on ARC-AGI-3 is noteworthy. There are currently no further developments.

HCLTech becomes a Global Premier Partner in the Claude Partner Network

HCLTech joins the Claude Partner Network as a Global Premier Partner. According to information published on the Claude Blog on October 21, 2026, HCLTech will bring Claude into its solutions for banks, hospitals, telecom and manufacturers, and will establish a dedicated Anthropic Business Unit and Center of Excellence. On the HCLSoftware product side, Claude will be embedded in HCL BigFix, HCL AppScan and Actian AI analyst; meanwhile, Claude and Claude Code will be integrated into the IT service platform AI Force, with accompanying policy guardrails and human review. There are currently no subsequent development facts.

Artificial Analysis updates the MLCR-AA medical long-context leaderboard

Artificial Analysis updated the MLCR-AA medical long-context reasoning leaderboard on October 21, 2026. The leaderboard shows that Claude Sonnet 5.5 (max with fallback) and GPT-6 Astra (max) tie for first place with a 41.7% pass rate, followed closely by GPT-6.1 Sol (max) at 38.3%, while Qwen3.8 2.4T A95B leads open-weight models at 30.6%. This update changes the prompt to a medical record reviewer role, requires conclusions and evidence to be answered in the same paragraph, switches scoring to a three-model review panel, and rechecks 60 expert reference answers, revising 15 of them. The leaderboard also shows that the medical record review task remains far from solved: no model exceeds a 42% pass rate, and the median model is below 15%; Anthropic models sweep the top 18 for completeness, with Claude Sonnet 5.5 leading at 64.8%, and OpenAI models sweep the top 27 for accuracy, with GPT-6.1 Sol leading at 96.7%.

Pigment completes $250 million financing at a valuation of $2.3 billion

French AI office tool company Pigment has completed a $250 million financing round at a valuation of $2.3 billion. According to Zhidongxi citing Bloomberg, the round was led by Scaleup Europe Fund, with Nvidia and other institutions participating. Founded in 2019, the company's annual recurring revenue has surpassed $100 million, with customers including Google, Anthropic, Uber, and Siemens, and it plans to develop and deploy agents on top of enterprises' existing ERP software. Pigment hopes to compete for enterprise customers currently using AI office products from vendors such as Microsoft, Oracle, and IBM; L'Oréal is launching an AI compensation planning tool based on it for use by 95,000 employees. A 36Kr·AI report also mentions the size, valuation, lead investor, and participating investors of this financing round, and states that its annual recurring revenue has doubled compared with last year. There is currently no information on subsequent developments.

Lindquist publishes paper tracing the origins of AI hallucinations

Benjamin Lindquist, an assistant professor in MIT's History Department, has published a new paper in Past and Present tracing the early origins of AI hallucinations. The paper points out that in the mid-20th century, engineers deliberately injected randomness into algorithms in order to imitate human creativity; early researchers valued neural networks precisely because they occasionally gave answers that deviated from expectations. Lindquist also stated that today institutions such as OpenAI and Anthropic suppress this kind of "weirdness" for legal and commercial reasons, and that there is an unavoidable trade-off between creativity and predictability. At present, this research has been reported through an MIT News interview, and there have been no further developments.

Claude models are officially available on Google Cloud

Claude Models on Google Cloud (Anthropic-hosted) are now generally available. According to information published on the Claude Blog on October 22, 2026, Google Cloud customers can use the full Claude platform capabilities through their existing IAM credentials, billing, and committed use discounts, with new features and betas launching on the same day as the native Claude platform. This version supports capabilities including Claude Managed Agents (beta), code execution, computer use, Agent skills, MCP connector, and Prompt caching, and provides the Claude Console; Claude Fable 5.1, Opus 5.5, Sonnet 5.5, and Haiku 5.5 are already available, with Claude Opus 5.5's Fast Mode delivering up to 2.5x faster output speed, at input and output prices of $8 and $40 per million tokens. Anthropic operates the service as a data processor, currently offering US and global endpoints; customers who require Google as a data processor or have regional data residency requirements can still choose the Google-hosted version. Testing Catalog reported the same launch news on October 23, 2026.

View story:Claude models are now generally available on Google Cloud

Ado Kukic says Claude Duet has become his daily go-to for non-coding tasks

Developer Ado Kukic commented that Claude Duet has become his daily go-to tool for handling non-coding matters. This assessment appeared in a report on Anthropic opening a research preview of Claude Duet: the app lets Claude work as a proactive agent, proactively bringing up matters that need attention, completing tasks the user assigns, continuing to work after the user leaves, and confirming with the user before acting on their behalf. Currently, this story has only the same occurrence evidence from the single source mentioned above, with no subsequent developments.

Artificial Analysis compares the medical long-text evaluation results of 14 models

On October 22, 2026, Artificial Analysis compared the medical long-text performance of 14 models on two evaluations, MLCR-AA and HealthBench Professional. The results show that Claude Sonnet 5.5, Opus 5.5, and Fable 5.1 swept the top three spots for MLCR-AA completeness, leading the best OpenAI model by about 19 to 23 points; Opus and Sonnet scored 71.4% and 70.8% respectively on HealthBench Professional raw scores, slightly above GPT-6 Astra's 69.7%. OpenAI models, meanwhile, took the top spots for MLCR-AA accuracy; Anthropic models' single responses were about 4 to 6 times as long as OpenAI's, with accuracy of 19% to 61%, versus 89% to 97% for OpenAI. GPT-6 Astra (max) tied with Claude Sonnet 5.5 for first place on MLCR-AA and held a slight lead on the HealthBench Professional length-adjusted score; Claude Opus 5.5, however, fell from first on raw score to sixth on the length-adjusted score.

Claude autonomously completes protein design project as an agent

On October 21, 2026, it was reported that Claude autonomously completed an entire round of a protein design project as an agent in Anthropic's Claude Science environment. According to the report, Claude selected the target site on a scorpion venom neurotoxin on its own, drove 12 protein design programs and set scoring thresholds, and selected 180 molecules from the 318,484 generated to send for physical synthesis; of the 211 designs tested by Adaptyv Bio and Twist Bioscience, 17 successfully bound the target toxin. The request was made by Citrini Research, a financial firm with no laboratory or protein expertise. The report also noted that Anthropic did not call it an antivenom, because the tests needed to support that claim have not yet been conducted. There are currently no subsequent developments.

Claude models are now generally available on Google Cloud

Claude Models on Google Cloud (Anthropic-hosted) are now generally available. According to information published on the Claude Blog on October 22, 2026, Google Cloud customers can use the full Claude platform capabilities through their existing IAM credentials, billing, and committed use discounts, with new features and betas launching on the same day as the native Claude platform. This version supports capabilities including Claude Managed Agents (beta), code execution, computer use, Agent skills, MCP connector, and Prompt caching, and provides the Claude Console; Claude Fable 5.1, Opus 5.5, Sonnet 5.5, and Haiku 5.5 are already available, with Claude Opus 5.5's Fast Mode delivering up to 2.5x faster output speed, at input and output prices of $8 and $40 per million tokens. Anthropic operates the service as a data processor, currently offering US and global endpoints; customers who require Google as a data processor or have regional data residency requirements can still choose the Google-hosted version. Testing Catalog reported the same launch news on October 23, 2026.

Claude web, desktop, and mobile will enable auto mode by default

Anthropic announced that Claude's web, desktop, and mobile versions will enable auto mode by default. According to Testing Catalog, the change will take effect next week on the Pro, Max, and Team plans; Claude will no longer ask users for approval step by step, and instead automatic checks will block risky operations, with users able to switch back to manual mode at any time. Sina Tech reported the same day that in this mode another AI judges whether each step is permitted, no longer asking users via a pop-up for each action; the report cited Anthropic testing showing that of 1,053 paid testers, only 13.6% stopped a dangerous command inserted midway, while the AI stopped 89%, and after more than 50 consecutive pop-up clicks the human interception rate fell to about 5%. There are currently no subsequent developments.

Karina Nguyen proposed that early evidence of AI risk should be taken seriously

On October 23, 2026, Karina Nguyen put forward the view that early evidence should be taken seriously before AI risk causes large-scale harm. In the text, she used Anthropic's practice as an example: at least since 2022, Logan Graham led early warnings about AI-enabled cyberattacks, but at the time no VC took it seriously; Anthropic also built sycophancy evaluations years before real ChatGPT-related incidents occurred, with Ethan Perez and others participating. Currently, this view was published by Karina Nguyen herself, has not yet been cross-verified by other independent sources, and no subsequent developments have been seen.

OpenClaw says cybersecurity is developing faster than AI this week

On October 23, 2026, OpenClaw reposted and commented on the "personal AI agents" weekly roundup compiled by McBain, and offered this judgment: the only industry changing faster than AI this week is cybersecurity. The weekly roundup covers Anthropic, OpenAI, Meta, OpenClaw, Perplexity, and others. OpenClaw's comment made the above comparison in a teasing tone; there is currently no further follow-up development or independent source cross-verification.

Alibaba Cloud Model Studio launches Token Plan subscription service

Alibaba Cloud Model Studio launches the Token Plan AI large-model subscription service, using Credits as the unified metering and deduction unit. According to a report on October 22, 2026, a single plan covers text, multimodal vision, image and video generation, and speech processing, and is compatible with both the OpenAI and Anthropic protocols as well as the Harness toolset; the personal edition is divided into three tiers—Lite, Standard, and Pro—and has a 7-day rolling usage limit, while the team edition supports seat allocation and member usage auditing. The service currently runs only in the China North 2 (Beijing) region, and dedicated API keys cannot be mixed with pay-as-you-go keys or domains. A report on October 23 further stated that the personal edition ranges from Lite at 39 yuan/month (11,500 Credits) to Pro at 499 yuan/month (180,000 Credits), and the team edition starts at 150 yuan/seat/month; Credits are dynamically deducted based on model type, token usage, thinking mode, and tool calls, and when a plan's quota is exhausted, usage packs are automatically deducted; if both are exhausted, the service is suspended and does not switch to pay-as-you-go billing, and it is an independent product from Coding Plan, with no migration between the two.

English translation is being prepared.

Chen Guangyu joins the Discovery Loop team

On October 21, 2026, 17-year-old AI researcher Chen Guangyu joined the Discovery Loop team founded by Jeff Dean and others. Previously, on October 15, he announced his departure from Moonshot AI; during his time at Moonshot AI, he participated in work including Attention Residuals, the underlying technology of Kimi K3, and KDA compute kernel acceleration. Reports say Discovery Loop was co-founded by Jeff Dean, Sanjay Ghemawat, Quoc Le and Oriol Vinyals, with the goal of building an AI research automation system that automatically proposes hypotheses, designs and runs experiments, and the company is seeking a new funding round at a valuation of about $50 billion. Other reports mention that Chen Guangyu had denied rumors of joining Anthropic and had plans to return to Carnegie Mellon to study.

OpenAI accelerates push for about $30 billion in financing

OpenAI is accelerating its push for about $30 billion in financing. According to Reuters, OpenAI is aiming to complete this funding round at a valuation of more than $1.4 trillion before Anthropic goes public, a valuation at least 64% higher than the $852 billion it reached in March. OpenAI hopes to lock in commitments before Anthropic files its S-1 and lists in mid-November, so as to avoid a direct comparison. Subsequent reports said existing investors such as SoftBank and Abu Dhabi's MGX are in talks to participate; OpenAI told investors that growth accelerated after the release of GPT-6 Astra, with annualized revenue already reaching $50 billion and expected to rise to $70 billion by the end of the year. The financing is still in progress.

English translation is being prepared.