All AI News

Anthropic News (firm Scan)

Introducing Claude Opus 5

Anthropic has released Claude Opus 5, a frontier-class model priced at half the cost of Claude Fable 5 while achieving state-of-the-art results on coding and knowledge work benchmarks including Frontier-Bench and GDPval-AA. On ARC-AGI 3, Opus 5 scores three times higher than the next-best model; on Zapier AutomationBench, its pass rate is approximately 1.5× competitors at equivalent cost. Early enterprise adopters report material gains in financial modeling (9 percentage points higher accuracy, 60% less time), due diligence (17% improvement), legal agent work (26% fewer tokens at comparable quality), and scientific research workflows. The model introduces configurable effort settings allowing organizations to trade off intelligence against speed and token cost, and becomes the default model on Claude Max.

3 minRead

Anthropic News (firm Scan)

The Anthropic Economic Index Connector

Anthropic has launched the Economic Index connector for Claude, enabling users to query its AI labor-market dataset directly through natural language in claude.ai. The connector requires no installation and works across all Claude models, allowing queries such as which occupations use AI most, how task automation has shifted over the past year, and regional usage patterns. The underlying dataset tracks how Claude is actually being used across the economy and has previously served researchers, journalists, and policymakers; the connector makes the same data accessible to any user. Anthropic notes the Index reflects Claude usage patterns rather than the broader labor market, and the full datasets remain publicly available on its website.

3 minRead

Anthropic News (firm Scan)

Donating another $20 million to Public First Action

Anthropic has donated an additional $20 million to Public First Action, a nonpartisan AI policy and public education organization, bringing its total commitment to $40 million. The donation is restricted to public education and policy work and cannot be used to influence candidate elections. Anthropic cites accelerating AI capabilities—including Claude Mythos Preview's discovery of thousands of high-severity software vulnerabilities across major operating systems and browsers—as evidence that governance frameworks must advance alongside technical progress. The company is advocating for mandatory model testing, independent evaluation, civil enforcement mechanisms, tighter export controls on advanced chips, and government authority to slow or block deployment of AI models posing catastrophic risk.

3 minRead

Anthropic News (firm Scan)

Supporting Ambitious External Research Through the Anthropic Economic Futures Research Fund

Anthropic is committing $200 million to the Economic Futures Research Fund to support external research on preparing society for AI-driven economic disruption. The fund will prioritize grants in the $5–30 million range across five areas: AI's impact on workers at the firm level, retraining and workforce transitions, modernizing income support for displaced workers, building worker ownership stakes in AI-driven growth, and generating evidence on public investments. Eligible applicants include accredited universities, independent research institutes, and nonprofits with field-experiment experience; individual applicants are excluded. The initiative represents a strategic shift from the prior Economic Futures program toward fewer, larger, higher-impact bets, with an emphasis on pilots scalable enough to inform policy before disruption arrives.

3 minRead

Anthropic News (firm Scan)July 20

Apply for Anthropic's AI for Science Rare Disease Research Grants

Anthropic has opened grant applications under its AI for Science program with a focused call on rare genetic diseases, offering accepted applicants up to $50,000 in Claude API credits over six months. The program runs two parallel tracks: one supporting basic science researchers and clinical consortia (including a partnership with the Monarch Initiative and its DisMech disease-classification library), and one targeting early-stage biotechs seeking to compress rare disease drug development timelines. Anthropic estimates AI can accelerate regulatory dossier assembly, therapeutic target selection, and cross-disease mechanism discovery—areas where current timelines run one to two years from confirmed diagnosis to available treatment. Applications are open through August 2, 2026, with outputs from the basic science track to be made publicly available via the Monarch Initiative.

3 minRead

Anthropic News (firm Scan)July 14

Introducing Claude for Teachers

Anthropic is launching Claude for Teachers, providing verified U.S. K-12 educators free access to premium Claude capabilities through June 30, 2027. The product integrates with Learning Commons to align lesson planning to academic standards across all 50 states and connects to nine third-party edtech platforms including MagicSchool, Canva Education, and ASSISTments. Key features include AI-assisted lesson differentiation, automated class data analysis, and scheduled task execution, with student data protected under FERPA-compliant terms and excluded from model training. The initiative is backed by partnerships with the Gates Foundation, Teach for America, and the American Federation of Teachers, with a planned efficacy pilot in Detroit Public Schools.

3 minRead

Anthropic News (firm Scan)

Anthropic commits $10 million to Canadian AI research

Anthropic is committing $10 million CAD to Canadian AI research institutions, partnering with eight organizations including Amii, Mila, the Vector Institute, CHEO, CAMH, Université Laval, the University of Saskatchewan, and the University of Toronto Data Sciences Institute. Funding will support research in AI safety, health outcomes, mental health, low-resource languages, and biomedical science, with additional API credits flowing to hundreds of Canadian startups via the Anthropic for Startups program. A companion data release from the March 2026 Anthropic Economic Index shows Canada ranks eighth globally in Claude.ai usage, with per-capita adoption more than four times what population size would predict. The announcement positions the investment within a broader geopolitical argument that democracies investing in AI now will shape its governance frameworks going forward.

3 minRead

Anthropic News (firm Scan)

Inviting Hard Questions

Anthropic has launched a public initiative called 'Hard Questions' to systematically collect and respond to societal concerns about AI, framing this transparency effort as central to its Public Benefit Corporation mission. The company has already surveyed 52,000 Americans via the Anthropic Public Record and interviewed 81,000 Claude users across 159 countries and 70 languages through its Anthropic Interviewer tool. Additional research includes in-person focus groups, anonymized real-world Claude usage data, and the creation of the Anthropic Institute to study AI's societal challenges. Anthropic is now inviting the public to submit their hardest questions on AI—covering jobs, society, science, and governance—and has committed to publicly tracking its actions and disclosing shortfalls against its stated goals.

3 minRead

Anthropic News (firm Scan)

UST is bringing Claude to physical AI

Anthropic and UST, a global technology and engineering services firm, have formed a partnership to embed Claude across UST's industrial and enterprise platforms. UST is deploying Claude Code inside its iDEC hardware validation pipeline, where the model reads chip schematics, generates regression tests, and compares live equipment data against digital twins — a closed-loop workflow UST says already cuts validation cycle times 50–70%, compressing four-day turnarounds to 48 hours. Beyond semiconductor and manufacturing environments, UST is integrating Claude into platforms serving healthcare (claims and care management), telecom (network operations and outage response), and banking (workflow automation and legacy core modernization). UST will train 20,000 engineers, architects, and consultants on Claude worldwide and joins the Claude Partner Network as a Global Premier Partner.

3 minRead

Anthropic News (firm Scan)

Ben Bernanke appointed to Anthropic's Long-Term Benefit Trust

Anthropic has appointed former Federal Reserve Chair Ben Bernanke to its Long-Term Benefit Trust (LTBT), the independent oversight body charged with holding Anthropic accountable to its mission of responsible AI development. Bernanke led the Fed from 2006 to 2014, guiding the U.S. economy through the 2008 financial crisis, and won the 2022 Nobel Prize in Economic Sciences for his research on banking and the Great Depression. The LTBT holds authority to appoint Anthropic board members, advises leadership on AI risk and societal impact, and its trustees hold no equity in the company. Bernanke's appointment is explicitly tied to Anthropic's focus on AI's macroeconomic effects, including impacts on workforces and economies globally.

3 minRead

Anthropic News (firm Scan)

Reflect With Claude

Anthropic has launched a beta feature called Reflect that lets Claude users track, visualize, and analyze their AI usage patterns over periods of 1 to 12 months. The dashboard surfaces topic breakdowns, usage timing, and task categories, and frames activity through a 4D AI Fluency Framework covering delegation, description, discernment, and diligence. Users can set quiet hours, schedule break reminders, and engage in structured reflection prompts designed to preserve human agency in AI-assisted workflows. The feature excludes incognito chats, health integration data, and source files from connected tools; insights are not used for model training or other secondary purposes. Anthropic developed the tool in collaboration with MIT Media Lab, Boston Children's Hospital's Digital Wellness Lab, and the Family Online Safety Institute.

3 minRead

Anthropic News (firm Scan)

Government of Alberta Uses Claude to Find and Fix Cybersecurity Vulnerabilities

The Government of Alberta deployed Claude Code (Opus and Sonnet models) to conduct a comprehensive cybersecurity review of its 1,280 applications and 3,400 code repositories across 27 provincial ministries. A team of approximately 50 AI agents scanned 466 million lines of code in 20 hours — work estimated to take 6.5 years by conventional methods — identifying vulnerabilities that traditional automated scanning tools missed. Beyond detection, Claude Code generated and tested patches, wrote missing test suites, and in some cases rebuilt legacy systems in modern languages; a 25-year-old Java subsidy portal that originally took five months to build was reconstructed in four to five days. Alberta has published technical white papers and is hosting an industry day in July to provide other governments with a replicable blueprint for addressing the technical debt and security gaps common across public-sector systems worldwide.

3 minRead

Anthropic News (firm Scan)

More details on Fable 5's cyber safeguards and our jailbreak framework

Anthropic has published detailed documentation on the cybersecurity safety classifiers deployed with Claude Fable 5 and introduced a proposed AI jailbreak severity framework developed with Glasswing partners. The classifier system divides cybersecurity activities into four categories—prohibited use, high-risk dual use, low-risk dual use, and benign use—each with distinct blocking or monitoring behaviors calibrated to the dual-use nature of security capabilities. Fable 5's safety margin is deliberately wider than prior models, accepting a higher false-positive rate on benign prompts to reduce harmful outputs. The jailbreak severity framework is positioned as a standards proposal intended to create consistent vocabulary between AI developers and governments when assessing jailbreak risk. Anthropic has opened a HackerOne program for researchers to submit Fable 5 cyber jailbreaks and is soliciting public feedback on the framework.

3 minRead

Anthropic News (firm Scan)July 1

Introducing Claude Sonnet 5

Anthropic has released Claude Sonnet 5, its most capable mid-tier model to date, designed specifically for agentic workflows including autonomous multi-step task execution, tool use, browser control, and code generation. The model closes the performance gap with Opus 4.8 while launching at introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31, 2026, rising to $3/$15 thereafter. Safety evaluations show Sonnet 5 has a lower rate of undesirable behaviors than its predecessor Sonnet 4.6, improved resistance to prompt injection, and reduced hallucination and sycophancy, though it trails the more capable Opus 4.8 on behavioral alignment metrics. Cybersecurity risk is assessed as low—Sonnet 5 cannot develop working exploits—and real-time cyber safeguards are enabled by default. The model is immediately available across all Claude plans, in Claude Code, and via API, with increased rate limits to support higher token usage at elevated effort levels.

3 minRead

Anthropic News (firm Scan)

Redeploying Fable 5

Anthropic's Claude Fable 5 and Mythos 5 models were subject to a U.S. government export control order from June 12–30 after Amazon researchers identified a technique to bypass Fable 5's safety classifiers, enabling it to identify software vulnerabilities and produce exploit demonstration code. Testing confirmed the same outputs were reproducible by numerous less-capable models, and the bypass did not expose Mythos-level offensive capabilities. Anthropic deployed an improved safety classifier that blocks the reported technique in over 99% of cases, validated by NIST's Center for AI Standards and Innovation. Fable 5 is restored globally as of July 1, Mythos 5 access is being expanded to approved U.S. organizations, and Anthropic is co-developing a shared industry jailbreak severity framework with Amazon, Microsoft, and Google alongside deeper pre-release government collaboration.

3 minRead

Anthropic News (firm Scan)June 30

Claude Science: An AI Workbench for Scientists

Anthropic has launched Claude Science, an AI workbench for scientific researchers available in beta for Pro, Max, Team, and Enterprise users. The platform integrates over 60 curated scientific tools and databases—spanning genomics, proteomics, single-cell analysis, structural biology, and cheminformatics—into a single environment that manages compute resources, produces fully auditable and reproducible artifacts, and coordinates multiple specialized agents. Early beta users report compressing two-year literature review timelines to weeks and accelerating complex genomic analyses to one-tenth of prior timelines. Claude Science runs on a lab's own infrastructure (local machines, HPC clusters, or on-demand GPUs via Modal), integrates with NVIDIA's BioNeMo Agent Toolkit, and includes a reviewer agent that checks citations, calculations, and figure fidelity in real time.

3 minRead

Anthropic News (firm Scan)

Introducing Claude Tag

Anthropic has launched Claude Tag, a Slack-native AI collaboration product available in beta for Enterprise and Team customers, allowing teams to tag @Claude in channels to delegate tasks asynchronously. Claude Tag is multiplayer—one Claude instance per channel, visible to all members—and builds persistent context over time across connected channels and data sources. Internally, 65% of Anthropic's product team code is now generated by the tool, with adoption spreading to functions including data analysis, support, and debugging. Administrators control tool and data access at the channel level, set token spend limits for the organization and individual channels, and retain a full audit log of all activity. The product runs on Claude Opus 4.8 and replaces the existing Claude in Slack app, with a 30-day migration window and introductory launch credits for eligible organizations.

3 minRead

Anthropic News (firm Scan)

Anthropic opens Seoul office and announces new partnerships across the Korean AI ecosystem

Anthropic has opened a Seoul office and announced a cluster of enterprise, startup, and academic partnerships across the Korean AI ecosystem. Major conglomerates are deploying Claude at scale: NAVER has rolled out Claude Code to its entire engineering organization, Samsung SDS is deploying Claude across Samsung Electronics, LG CNS is extending access across LG Group, and Hanwha Solutions is using AWS Bedrock to meet data-residency requirements. Channel Corp embeds Claude in its customer AI platform serving over 230,000 businesses across Korea, Japan, and the United States. Anthropic will also provide Claude access to up to 60 researchers through the National AI Research Lab consortium, spanning KAIST, Korea University, Yonsei University, and POSTECH.

3 minRead

Anthropic News (firm Scan)

Statement on the US government directive to suspend access to Fable 5 and Mythos 5

The US government has issued an export control directive requiring Anthropic to immediately suspend all access to its Fable 5 and Mythos 5 models for all customers globally, citing national security concerns over a reported jailbreak method. Anthropic is complying but publicly disputes the basis for the directive, stating the demonstrated technique constitutes only a narrow, non-universal jailbreak that produces results already achievable by other publicly available models including GPT-5.5. The company argues that applying this recall standard industry-wide would effectively halt all frontier model deployments. Anthropic's defense-in-depth strategy for Fable 5 included thousands of hours of government and third-party red-teaming, 30-day mandatory data retention for jailbreak monitoring, and safeguards benchmarked as stronger than any previously deployed model. All other Anthropic models remain available, and the company states it is working to restore Fable 5 and Mythos 5 access as quickly as possible.

3 minRead

Anthropic News (firm Scan)

DXC will integrate Claude into the systems banks, airlines, and other regulated industries rely on

Anthropic and DXC Technology have announced a multi-year global alliance in which DXC will deploy Claude across the mission-critical IT systems it operates for major banks, airlines, insurers, manufacturers, and government agencies. DXC will certify tens of thousands of forward-deployed engineers through Anthropic Academy to embed Claude directly inside client environments under regulated, compliance-heavy conditions. DXC validated the model in its own 115,000-person operations first, using Claude to generate more than 95% of the code for OASIS, its new AI-native managed-services platform, while achieving a claimed 10x acceleration in software development. Initial deployment focus areas include insurance modernization, legacy codebase refactoring, cybersecurity SOC operations, and application maintenance. DXC has also joined the Claude Partner Network, extending Anthropic's reach into large-enterprise regulated verticals at scale.

3 minRead

Anthropic News (firm Scan)

TCS and Anthropic Partner to Bring Claude to Regulated Industries

Anthropic and Tata Consultancy Services (TCS) have announced a partnership to deploy Claude across regulated industries globally. TCS will roll out Claude to 50,000 of its own employees across 56 countries and build Claude-powered, industry-specific products for clients in financial services, healthcare, public sector, life sciences, aviation, telecom, and medical technology. Concrete deployments are already underway: Diligenta will use Claude to serve 22 million UK life and pensions policyholders, TCS banking teams will use Claude Code for software engineering productivity, and TCS iON will deliver Claude training across 75 million annual assessments in India. TCS joins the Claude Partner Network as a systems integrator and implementation partner, packaging Claude into vertical offerings such as claims adjudication and lending advisory, with Anthropic citing India as its second-largest market.

3 minRead

Anthropic News (firm Scan)

Introducing Claude Corps

Anthropic is launching Claude Corps, a national fellowship program backed by an initial $150 million commitment that will place 1,000 paid fellows at nonprofits across the United States. Fellows receive a $85,000 annual salary, benefits, and ongoing AI training, and are employed through CodePath while spending 12 months full-time embedded at host organizations. The program targets at least 400 nonprofits in its first year, spanning food banks, veteran services, workforce development, and conservation. Anthropic frames the initiative as both a direct investment in workers absorbing AI-driven economic disruption and a scalable model for broadening AI's benefits, with Social Finance leading measurement and a longer-term funding vehicle to enable expansion.

3 minRead

Anthropic News (firm Scan)June 9

Claude Fable 5 and Claude Mythos 5

Anthropic has launched Claude Fable 5 for general use and Claude Mythos 5 for a restricted set of cyberdefense and infrastructure partners, both priced at $10 per million input tokens and $50 per million output tokens—less than half the cost of Claude Mythos Preview. Fable 5 leads on nearly all tested benchmarks, with documented enterprise results including Stripe compressing two months of Ruby codebase migration work into a single day and IMC reporting near-perfect scores on trading-analysis evaluations. Mythos 5, deployed initially through Project Glasswing in collaboration with the US government, carries lifted cybersecurity safeguards and demonstrated a roughly 10x acceleration in internal drug design workflows, with 9 of 14 protein targets yielding viable drug candidates. Fable 5 ships with conservative safety filters that redirect flagged queries to Opus 4.8, triggering in fewer than 5% of sessions on average, while alignment assessments show misaligned behavior at levels comparable to Opus 4.8. Anthropic intends to expand Mythos 5 access through a broader trusted access program and plans to publish novel genomics research conducted autonomously by the model.

3 minRead

Anthropic News (firm Scan)

Election Safeguards Update

Anthropic has published a detailed update on the election safeguards built into its Claude models ahead of the 2026 US midterms and other major global elections. Claude Opus 4.7 and Sonnet 4.6 scored 95–96% on political impartiality evaluations and responded appropriately to election-related prompts 100% and 99.8% of the time, respectively, across 600 test cases covering both harmful and legitimate requests. The company tested for the first time whether models could autonomously run influence operations end-to-end; with safeguards active, both new models refused nearly every task, though results without safeguards signal the need for continued vigilance. Anthropic is deploying election banners on Claude.ai directing users to nonpartisan voting resources (TurboVote for the US, with Brazil to follow), and web search triggers on election-related queries at a 92–95% rate to surface real-time information. Third-party review partnerships with organizations including the Future of Free Speech at Vanderbilt and the Collective Intelligence Project are underway to independently validate model behavior around political expression.

3 minRead

Anthropic News (firm Scan)June 3

Introducing the Services Track and Partner Hub of the Claude Partner Network

Anthropic has expanded its Claude Partner Network with two new structures: a tiered Services Track and a Claude Partner Hub portal. The Services Track classifies partner firms across three tiers—Select (10+ certified individuals, 2+ production deployments), Preferred (100+ certified, 15+ deployments), and Global Premier (1,000+ certified, 100+ deployments across 3+ regions)—with promotions reviewed twice annually and demotion only after 90 days' notice. The Partner Hub gives firms a daily-refreshed dashboard of their standing and provides enterprises a public directory to evaluate partners by certified headcount, production deployments, and published customer references. Major professional-services firms are already embedded in the network: Deloitte (470,000 employees with access), Cognizant (350,000), PwC (global rollout in progress), KPMG (276,000+), Accenture (30,000 trained), and Infosys building industry-specific agents. The $100 million commitment announced in March funds partner training, technical support, and co-marketing, with more than 10,000 individual Claude certifications already earned.

3 minRead

Anthropic News (firm Scan)

What we learned mapping a year's worth of AI-enabled cyber threats

Anthropic analyzed 832 accounts banned for malicious cyber activity between March 2025 and March 2026, mapping attacker behavior to the MITRE ATT&CK framework. Three findings stand out: AI is enabling less-skilled actors to execute advanced post-compromise techniques like lateral movement and privilege escalation, previously reserved for sophisticated attackers; the share of actors classified as medium risk or higher jumped from 33% to 56% across the two six-month periods studied; and traditional risk signals—number of techniques used, platform choice—no longer reliably distinguish low- from high-risk actors. The most dangerous actors build agentic architectures that chain attack stages autonomously with minimal human input, a behavior category not yet captured in the MITRE ATT&CK framework. Anthropic is in discussions with MITRE to update the framework and has deployed model-level safeguards targeting observed malicious behaviors such as malware development and mass data exfiltration.

3 minRead

Anthropic News (firm Scan)

Expanding Project Glasswing

Anthropic is expanding Project Glasswing, its AI-powered cybersecurity initiative, from roughly 50 initial partners to approximately 150 additional organizations across 15+ countries. Using Claude Mythos Preview, early partners have already identified more than 10,000 high- or critical-severity vulnerabilities in their codebases. The expanded cohort covers previously underrepresented critical infrastructure sectors—power, water, healthcare, communications, and hardware—where a successful cyberattack could affect more than 100 million people per partner. Anthropic warns that Mythos-class AI cyber capabilities will be broadly available across the industry within 6 to 12 months, and is working to develop robust safeguards enabling general access while simultaneously scaling vulnerability patching, disclosure workflows, and a Cyber Verification Program for trusted organizations.

3 minRead

Anthropic News (firm Scan)

Anthropic Confidentially Submits Draft S-1 to the SEC

Anthropic, PBC has confidentially submitted a draft Form S-1 registration statement to the U.S. Securities and Exchange Commission, formally initiating the process for a potential initial public offering of its common stock. The filing preserves optionality; a public offering will proceed only after SEC review and subject to market conditions. Share count and offering price have not been determined. The announcement follows Anthropic's Series H raise of $65 billion at a $965 billion post-money valuation, positioning it as one of the most highly valued private technology companies ahead of a potential listing.

3 minRead

Anthropic News (firm Scan)

Introducing Claude Opus 4.8

Anthropic has released Claude Opus 4.8, an upgrade to its flagship model available at the same price as its predecessor. Key capability improvements include benchmark gains across coding, agentic tasks, and reasoning, with the model being approximately 4 times less likely to allow code flaws to pass unremarked compared to Opus 4.7. Fast mode now runs at 2.5× speed at 3× lower cost than prior models, and Databricks reports a 61% reduction in token cost versus Opus 4.7 for multimodal workloads. Accompanying features include dynamic workflows in Claude Code enabling hundreds of parallel subagents for large-scale engineering tasks, user-controlled effort levels, and a mid-task system instruction API that preserves prompt cache integrity. Alignment evaluations show Opus 4.8 has substantially lower rates of deceptive or misuse-enabling behavior than Opus 4.7, comparable to Anthropic's best-aligned model.

3 minRead

Anthropic News (firm Scan)

Anthropic opens Milan office to support Italian enterprise, research, and developers

Anthropic is opening its sixth European office in Milan, expanding its footprint alongside London, Dublin, Paris, Zurich, and Munich. The Milan team, led by Head of Southern Europe Thomas Remy, is already engaged with major Italian enterprises across finance (Generali Group, Unipol Group), life sciences (Angelini Pharma, Bracco Group), energy (Enel Group), and automotive (Pirelli). Early deployment results are measurable: JAKALA rolled out Claude across 3,000+ seats, recovering roughly 70% of senior staff time for higher-value work; Satispay compressed an 18-month engineering roadmap to seven months and accelerated core payment system updates tenfold. Bending Spoons reports that the majority of its code changes are now co-authored with Claude Code, illustrating the depth of AI integration already underway in Italian tech.

3 minRead

Anthropic News (firm Scan)

Anthropic raises $65B in Series H funding at $965B post-money valuation

Anthropic has closed a $65 billion Series H round at a $965 billion post-money valuation, led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. Run-rate revenue crossed $47 billion earlier this month, up from the Series G close in February 2025. The raise includes $15 billion in previously committed hyperscaler investments and adds Micron, Samsung, and SK hynix as strategic infrastructure partners. Proceeds will fund safety and interpretability research, compute expansion, and product scaling; Anthropic has already secured agreements for up to 10 gigawatts of new compute capacity across Amazon, Google/Broadcom, and SpaceX. Claude is now available on all three major cloud platforms—AWS, Google Cloud, and Microsoft Azure—with AWS remaining the primary cloud and training partner.

3 minRead

Anthropic News (firm Scan)

Anthropic appoints KiYoung Choi as Representative Director of Korea

Anthropic has appointed KiYoung Choi as Representative Director of Korea, ahead of opening its Seoul office. Korea ranks as one of Anthropic's highest-engagement markets globally, with Claude.ai usage running 3.5x above the expected rate for its population size, skewed toward technical and creative applications. Choi brings 30+ years of technology leadership across Korea and Asia-Pacific, with prior country-level roles at Snowflake, Google Cloud, Adobe, Autodesk, and Microsoft. The Korea team will target enterprise and startup partnerships, government and research engagement, and developer community support, building on existing deployments with customers such as SK Telecom and Law&Company.

3 minRead

Anthropic News (firm Scan)May 25

Anthropic co-founder Chris Olah's remarks on Pope Leo XIV's encyclical "Magnifica humanitas"

Anthropic co-founder Chris Olah delivered remarks at the Vatican presentation of Pope Leo XIV's May 2026 encyclical on AI, 'Magnifica humanitas.' Olah acknowledged that frontier AI labs, including Anthropic, operate under commercial, geopolitical, and competitive pressures that can conflict with responsible development, making external moral and institutional voices essential. He identified three priority areas for non-technical discernment: equitable distribution of AI's economic gains to the global poor, frameworks for human and family flourishing amid labor displacement, and the unresolved philosophical and ethical questions posed by AI systems that exhibit internal states resembling emotion and introspection. Olah framed the Vatican's engagement as a model for the broader civil society, religious, and governmental scrutiny that AI developers require but cannot generate from within.

3 minRead

Anthropic News (firm Scan)

Announcing Our Updated Responsible Scaling Policy

Anthropic has published a major update to its Responsible Scaling Policy (RSP), the risk governance framework governing when and how it trains and deploys frontier AI models. The updated policy introduces two explicit Capability Thresholds that trigger mandatory safeguard upgrades: autonomous AI R&D acceleration (requiring ASL-4 or higher standards) and meaningful uplift for CBRN weapons creation (requiring ASL-3 standards). All current Anthropic models operate under ASL-2 standards, reflecting current industry best practices. The update adds structured capability and safeguard assessments, safety-case-style documentation processes, internal stress-testing, and external expert review, while a compliance audit of the first year identified minor procedural gaps that posed no material safety risk. Jared Kaplan succeeds Sam McCandlish as Responsible Scaling Officer, and Anthropic is hiring a Head of Responsible Scaling to coordinate cross-company RSP execution.

3 minRead

Anthropic News (firm Scan)

Anthropic Acquires Stainless

Anthropic has acquired Stainless, a 2022-founded company specializing in SDK generation and MCP server tooling. Stainless has built every official Anthropic SDK since the API's launch and serves hundreds of companies with auto-generated SDKs across TypeScript, Python, Go, Java, and other languages. The acquisition is strategically tied to Anthropic's Model Context Protocol (MCP), which the company created to standardize agent connectivity to external data and tools. By internalizing Stainless, Anthropic aims to deepen Claude's ability to integrate with third-party systems as AI deployments shift from single-turn model queries to multi-step autonomous agents.

3 minRead

Anthropic News (firm Scan)

KPMG integrates Claude across its core business and workforce of more than 276,000 in strategic alliance

KPMG has formed a global strategic alliance with Anthropic to deploy Claude across its entire workforce of 276,000+ employees in 138 countries, with immediate integration into Digital Gateway, KPMG's core client-facing platform built on Microsoft Azure. Claude is being embedded directly into tax and legal workflows via Claude Cowork and Managed Agents, reducing AI agent build times from weeks to minutes. Anthropic is naming KPMG a preferred partner for private equity, enabling KPMG to deploy Claude into PE portfolio companies through offerings including KPMG Blaze, which integrates Claude Code to accelerate IT modernization. The alliance extends to cybersecurity vulnerability detection and remediation, guided by KPMG's Trusted AI framework, with joint research from UT Austin's McCombs School of Business informing responsible human-in-the-loop deployment practices.

3 minRead

Anthropic News (firm Scan)

Introducing Claude Design by Anthropic Labs

Anthropic has launched Claude Design, an AI-powered visual collaboration tool available in research preview to Claude Pro, Max, Team, and Enterprise subscribers. Powered by Claude Opus 4.7, the product enables users to generate and iterate on designs, prototypes, pitch decks, wireframes, and marketing assets through natural language conversation and fine-grained controls. Enterprise teams can onboard their existing design systems, codebases, and brand assets so all output remains on-brand by default; finished work exports to Canva, PPTX, PDF, or standalone HTML and can be handed off directly to Claude Code for implementation. Early adopters including Datadog and Brilliant report compressing multi-day design-review cycles into single conversations, with complex prototypes requiring as few as 2 prompts versus 20+ in prior tools.

3 minRead

Anthropic News (firm Scan)

Introducing Claude for Small Business

Anthropic launched Claude for Small Business, a product targeting the 44% of U.S. GDP and nearly half the private-sector workforce represented by small businesses. The offering integrates with QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365 via toggle installation inside Claude Cowork, delivering 15 pre-built agentic workflows and 15 task-level skills across finance, operations, sales, marketing, HR, and customer service. Key workflows include payroll planning, month-end close, cash-flow dashboards, invoice chasing, and campaign execution—all requiring user approval before any action is taken. The launch includes a free AI Fluency for Small Business course in partnership with PayPal and a 10-city hands-on workshop tour beginning May 14 in Chicago.

3 minRead

Anthropic News (firm Scan)

Claude Is a Space to Think

Anthropic has committed to keeping Claude permanently free of advertising, citing the incompatibility of ad-based incentives with its core design principle of acting unambiguously in users' interests. The company's analysis of Claude conversations found a significant share involve sensitive, personal, or complex professional topics where commercial influence would be inappropriate and potentially harmful. Anthropic's revenue model relies exclusively on enterprise contracts and paid subscriptions, which it argues aligns AI recommendations with user benefit rather than advertiser goals. The piece distinguishes future commerce features—such as agentic purchasing initiated by users—from advertising, framing the dividing line as whether the AI is working for the user or for a third-party commercial interest. Anthropic acknowledges the tradeoffs of this model and commits to transparency if the approach changes.

3 minRead

Anthropic News (firm Scan)

Agents for Financial Services

Anthropic has released ten ready-to-run agent templates targeting high-volume financial services workflows, including pitchbook construction, KYC screening, general ledger reconciliation, month-end close, and statement auditing. The templates deploy as plugins in Claude Cowork or Claude Code, or as autonomous Claude Managed Agents with audit logs, credentialed data access, and long-running session support. Claude now integrates directly with Microsoft Excel, PowerPoint, Word, and Outlook via add-ins, with context persisting across applications so work initiated in a financial model can flow into a presentation without manual re-entry. Eight new data connectors—including Dun & Bradstreet, SS&C Intralinks, Verisk, and a Moody's MCP app covering 600 million entities—expand the governed data ecosystem, and Claude Opus 4.7 leads Vals AI's Finance Agent benchmark at 64.37%.

3 minRead

Anthropic News (firm Scan)

Gates Foundation Partnership

Anthropic and the Gates Foundation are committing $200 million over four years in grant funding, Claude usage credits, and technical support across global health, life sciences, education, and economic mobility. In global health, the partnership targets improved outcomes for the 4.6 billion people lacking access to essential health services, including AI-accelerated vaccine and therapy development for polio, HPV, and eclampsia, and AI-enhanced disease-transmission forecasting for malaria and tuberculosis. In education, the two organizations will co-develop AI tutoring and literacy tools for K-12 students in the US, sub-Saharan Africa, and India, releasing public benchmarks and datasets. The economic mobility workstream covers agricultural productivity tools for smallholder farmers, portable skills records, and career guidance platforms in the US. Anthropic's Beneficial Deployments team leads this initiative, which also encompasses discounted Claude access for nonprofits and development of AI public goods including health datasets and evaluation benchmarks.

3 minRead

Anthropic News (firm Scan)

Higher usage limits for Claude and a compute deal with SpaceX

Anthropic has signed a compute agreement with SpaceX granting access to 300+ megawatts of capacity (220,000+ NVIDIA GPUs) at the Colossus 1 data center, effective immediately. This deal joins a portfolio of infrastructure commitments totaling over 10 GW across Amazon (up to 5 GW, with ~1 GW online by end of 2026), Google and Broadcom (5 GW, online 2027), a $30 billion Microsoft/NVIDIA Azure deal, and a $50 billion Fluidstack infrastructure investment. As a direct result of expanded capacity, Anthropic is doubling Claude Code's five-hour rate limits for Pro, Max, Team, and Enterprise plans, removing peak-hour throttling for Pro and Max, and significantly raising API rate limits for Claude Opus models. International capacity expansion is underway with a focus on data residency compliance for regulated industries including financial services, healthcare, and government.

3 minRead

Anthropic News (firm Scan)

PwC Expanded Partnership

Anthropic and PwC have expanded their strategic alliance, with PwC deploying Claude across its global workforce of hundreds of thousands of professionals to build agentic technology, execute deals, and reinvent enterprise functions for clients. PwC is launching a dedicated Office of the CFO business group—its first standalone unit anchored in Anthropic technology—targeting regulated industries where accuracy and auditability are critical, including banking, insurance, and healthcare. The partnership includes a joint Center of Excellence and certification of 30,000 U.S. professionals on Claude, with Claude Code and Claude Cowork rolling out firm-wide. Production deployments are already delivering measurable results: insurance underwriting cycles cut from 10 weeks to 10 days, cybersecurity incident response reduced from hours to minutes, and overall delivery time improvements of up to 70% across live client engagements. Anthropic's $100 million Claude Partner Network investment underpins this collaboration, which it describes as the deepest commitment within that program.

3 minRead

Anthropic News (firm Scan)

Widening the conversation on frontier AI

Anthropic is expanding its AI development process by conducting structured dialogues with scholars, clergy, philosophers, and ethicists from more than 15 religious and cross-cultural traditions to inform the moral formation of its Claude models. The initiative focuses on how AI character should be shaped, drawing on centuries of accumulated thinking about virtue, ethics, and human development rather than aligning to any single worldview. One concrete outcome already emerged: an experiment giving Claude a mid-task tool that surfaces its own ethical commitments produced measurably lower rates of misaligned behavior on internal alignment evaluations, with Anthropic still isolating whether the effect stems from the reminder content or the act of pausing to reflect. Future conversations will extend to legal scholars, psychologists, writers, and civic institutions, with the scope broadening to AI's effects on work, institutions, and power distribution. The effort is positioned to directly influence Claude's constitution, training values, and behavioral evaluations.

3 minRead

Anthropic News (firm Scan)May 4

Introducing Claude Opus 4.7

Anthropic has released Claude Opus 4.7, now generally available across Claude products, API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry at $5 per million input tokens and $25 per million output tokens—same pricing as Opus 4.6. The model delivers measurable gains in advanced software engineering, with early-access partners reporting 13% higher coding task resolution, 3x more production task completions versus Opus 4.6, and a 70% pass rate on CursorBench versus 58% for its predecessor. Opus 4.7 introduces higher-resolution vision, stronger instruction-following, and self-verification of outputs—capabilities cited as enabling hands-off execution of complex, long-running agentic workflows. The release also marks Anthropic's first deployment of real-time cybersecurity safeguards under its Project Glasswing framework, with a Cyber Verification Program for legitimate security professionals, as the company tests guardrails ahead of a broader release of its more capable Mythos-class models.

3 minRead

Anthropic News (firm Scan)December 11

Results from First Anthropic Public Record

Anthropic surveyed 51,993 Americans in November–December 2025 to establish a public opinion baseline on AI attitudes. Job displacement is the top fear across all demographics at 64%, followed by cognitive dependency (56%) and misinformation (52%), while 48% of respondents ranked curing diseases as their primary hope for AI. Only 15% of Americans trust AI companies to self-govern development decisions, and over 70% support government regulation — with bipartisan backing for action on privacy, child safety, and liability. Notably, AI anxiety does not follow typical partisan or educational divides in direction, only in intensity, and daily AI users are significantly less worried about job loss (54%) than non-users (70%).

3 minRead