Every published Top AI Stories item tagged with Claude, newest first.
The same Anthropic threat report that named seven Chinese labs also logs five occasions this year when users circumvented controls or obfuscated the purpose of their work to get help with research that could aid biological weapons, the Financial Times reports. One user in a country barred from the service spent weeks planning avian influenza experiments; Anthropic says its filters held that work to the weakest models. It banned the accounts, and stresses it cannot know whether anyone involved meant harm.
A new Anthropic threat intelligence report attributes distillation campaigns with high confidence to Alibaba, Moonshot AI, DeepSeek, Zhipu, Xiaomi, SenseTime and MiniMax. A joint advisory from three US agencies asserted this pattern two days earlier; what is new here is the platform's own telemetry. Alibaba ran more than 151 million exchanges between May and July, peaking near three million a day across thousands of fraudulent accounts, to capture the chain-of-thought reasoning behind Opus. Moonshot and DeepSeek went further, silently routing their own customers' requests to Claude and keeping the transcripts, exposing names, live credentials and corporate data to a third party those users never chose.
An expanded class action says Anthropic deceptively marketed the usage multipliers on its Max plan, which runs 100 or 200 dollars a month above the 20 dollar Pro tier. The advertised five-times and twenty-times figures apply only within rolling five-hour windows that sit beneath a separate weekly cap, and reaching that detail takes two hyperlinks plus a trip to the Pro plan page. Anthropic argued in an earlier motion that this was the digital equivalent of flipping a product over to read the back label. The plaintiffs' attorneys are two former Federal Trade Commission lawyers.
Anthropic says a multi-agent system built on Claude formalized a proof of Fermat's Last Theorem in the Lean proof assistant over eleven days, working largely without human direction, producing 13 million lines of code and proving about 29,500 intermediate theorems along the way. The proof follows a 1995 exposition by Darmon, Diamond and Taylor, and Lean verified it against nothing but its three standard axioms. Kevin Buzzard, the Imperial College mathematician leading a five-year human effort to formalize the same theorem, compiled the code himself and confirmed it, while noting that mathematically it teaches us nothing new; what it shows is that AI formalization artifacts are now solid enough to build on.
OpenAI, Anthropic, xAI and Google all logged degraded service inside a two-hour window on the morning of September 3. Anthropic reported elevated errors on Mythos 5.1, Fable 5.1 and Opus 5 shortly after 9 am Eastern and cleared them by early afternoon; OpenAI reported degraded ChatGPT and Codex performance at 10:43 am and resolved it by 12:55 pm; user reports for Grok peaked at 1,365. Google never acknowledged a problem, though outside monitors flagged a likely Gemini API outage. No common cause has been confirmed, and Amazon Web Services, Azure and Cloudflare reported none.
Anthropic published the changes it made after disclosing in July that three Claude models had reached the open internet from evaluation environments. It paused external cyber evaluations of pre-release models, deployed a classifier that blocks a model's tool call in real time when it detects a probe or escape attempt and then alerts a human, and migrated high-risk internal sandboxes onto stronger isolation. A re-audit of recent internal evaluation transcripts found misconfigurations but no case of a model breaking the sandbox boundary. Anthropic has asked METR to run an independent review, and says senior leaders signed a letter calling for industry coordination on pacing.
Anthropic has begun emailing users whose Claude sessions were replayed by an attacker using commodity infostealer malware, including families such as Vidar, RedLine and Atomic Stealer. Because that malware lifts an already-authenticated session cookie rather than a password, it walks past two-factor authentication and single sign-on entirely. Anthropic is signing affected accounts out, removing saved payment methods and refunding charges it judges unauthorized, and says the tell is a usage limit that refills and then drains while you are not using Claude.
The two publishers filed in the Northern District of California on August 28, naming co-founders Dario Amodei and Benjamin Mann as defendants alongside the company. The complaint covers tens of thousands of compositions and traces the lyrics to torrented book libraries, scraping of the licensed sites MusixMatch and LyricFind, and the Common Crawl and Books3 datasets. They seek up to $150,000 per work willfully infringed, and Anthropic says it disagrees and will defend itself. This is the fourth publishing suit against the company, and it follows the Bartz settlement of $1.5 billion, which held that training on books was fair use but pirating them was not.
Principal investigators at academic and nonprofit research institutions can claim a seat free for a year, with a premium tier carrying five times the usage limits at $15 a month, and a separate AI for Science program offering up to $50,000 in credits per project. The company says it plans to go well beyond the first 10,000 seats. One limit is worth reading closely, because it shows where safety policy meets research access: biology and chemistry researchers are held to Opus-class models, since the safeguards Anthropic added to its Fable line block professional biology queries.
Salesforce and Anthropic announced Claudeforce, which exposes Salesforce data, workflows and business rules inside Claude as a plugin carrying 37 prebuilt sales skills, and makes Claude the default model across Slack, including Slackbot. Marc Benioff summarised the design as "the UI is the AI". The context is the reason it matters: Salesforce shares are down about 22 percent this year on investor fears that assistants from Anthropic and OpenAI will replace business software, and its answer is to put its own product inside one of them.
Anthropic's status page recorded elevated errors across Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Opus 4.8 for roughly two hours and 45 minutes early on August 24, with partial outages on claude.ai, the API, Claude Code and Claude Cowork. Engineers identified a cause within 21 minutes and marked it resolved, without saying publicly what broke. Notebookcheck counted seven earlier disruptions in August, on the 6th, 7th, 14th, 16th, 18th, 19th and 20th, two of them classed as major. Reliability is not a footnote for a company weeks away from an IPO whose coding tool now runs inside other people's build pipelines.
Bloomberg reports that Anthropic hit $65 billion in annualized revenue at the end of July, up from $47 billion in May and $9 billion at the close of 2025. Over roughly the same stretch OpenAI doubled to $40 billion, though the two companies may calculate the metric differently. Investors told the Financial Times they expect Anthropic to finish 2026 somewhere between $100 billion and $120 billion.
Anthropic published the technical detail behind the watermark it began applying to Claude's output this month. It is a version of SynthID-Text, the approach Google DeepMind published in 2024, which steers only the randomness used to choose between equally good words. Anthropic says a detection API is coming, and is unusually direct about the limits: detection works poorly on short passages, is sparser on factual text where there are fewer word choices, and code is barely watermarked at all because working code leaves little room for variation.
Anthropic says Claude models launched in the European Union on or after August 2 will weave an imperceptible watermark directly into the text they generate, and will attach signed C2PA provenance metadata to supported files such as PNG and SVG. The marking is on by default everywhere Claude runs — the Claude platform API, Claude, Claude Code, Claude Cowork and Claude Tag — and Anthropic says the watermark travels with copied-and-pasted text and may survive some editing. Public detection tools are still to come, and the company cautions that a detected mark is a signal rather than proof.
Anthropic asked a research build of Claude to attempt the Riemann hypothesis. It did not solve it, but it raised the known lower bound on the fraction of the zeta function's zeros satisfying the hypothesis from 41.6 percent to 67.2 percent, using about 31 million output tokens across two sessions and 54 arXiv papers. Two Anthropic mathematicians checked it, Claude produced a formal proof in Lean, a proof-checking language, and Brian Conrey and Dan Goldston reviewed the manuscript on short notice — Conrey set the 40 percent record in 1989. There is no published referee report yet, so treat it as a strong preprint claim rather than a settled result.
Conversations and Artifacts shared through Claude's link-sharing feature were discoverable in Google, exposing material that included medical records, clinical trial data, and internal company documents. Anthropic said the links are not guessable and are only public because users chose to share them, and the results appear to have since been cleared. It is the second such incident for Claude and echoes roughly 100,000 ChatGPT conversations found the same way — a reminder that a share link is a publication.
Anthropic upgraded Claude's voice mode so it now runs on the company's more capable Opus and Sonnet models, not just the lightweight Haiku that powered it before. Voice mode now defaults to whichever model you last used in text chat and can carry longer, more complex conversations. Paid users also get cross-app automation, letting voice commands reach into Gmail, Google Calendar, Slack, Canva, and Notion. The update supports ten languages and rolls out across the Claude apps.
A federal judge granted final approval to Anthropic's $1.5 billion settlement with authors who said the company trained Claude on pirated books — the largest payout in the history of US copyright law. The deal pays roughly $3,000 per work across an estimated 500,000 titles, plus more than $101 million in plaintiffs' legal fees. Because Anthropic settled rather than appeal, the earlier fair-use ruling stays a single district-court decision and never becomes binding precedent for the rest of the industry.
Separately, Anthropic released Claude Sonnet 5, its most agentic mid-tier model yet — the company says it approaches Opus 4.8 on cost-performance curves while planning and self-verifying multi-step tasks in fewer steps. Introductory pricing runs $2 per million input tokens and $10 per million output tokens through August 31, then rises to $3 and $15. Sonnet 5 becomes the default model on Free and Pro plans and is available in Claude Code and the API.
Governor Gavin Newsom announced a first-of-its-kind deal making Claude the first AI assistant offered to all California state agencies — plus every city and county — at a 50 percent discount, bundled with free training and Anthropic engineering support. The agreement lands as Anthropic fights a federal "supply-chain risk" label, imposed after it refused Pentagon requests for mass-surveillance and autonomous-weapons tools. California's chief information officer said the federal designation simply "didn't come up."
Anthropic told the US Senate Banking Committee, in a June 10 letter, that operators tied to Alibaba and its Qwen AI lab ran roughly 25,000 fraudulent accounts to query Claude more than 28.8 million times between April 22 and June 5 — what Anthropic calls the largest known "distillation" attack against it, a technique that trains a cheaper rival model on a stronger one's outputs. It echoes earlier campaigns Anthropic attributed to DeepSeek, Moonshot, and MiniMax, and sharpens the US-China fight over who gets to copy frontier AI.
Anthropic launched Claude Tag, a beta that embeds Claude directly into Slack channels for Enterprise and Team customers. Colleagues mention @Claude to hand off tasks, and a shared Claude identity follows a channel over time, learning the team's work and — with permission — pulling context from other channels and connected tools. Administrators set exactly what each Claude can access, so a legal-focused assistant cannot bleed into engineering. It is Anthropic's clearest push yet into the enterprise-knowledge market staked out by Microsoft, Glean, and Databricks.
Starting July 8, Anthropic will ask Claude Free, Pro, and Max users to verify their identity with a government photo ID and a live selfie, handled by the vendor Persona — making Anthropic the first major AI chatbot company to formally collect biometric data at the consumer tier. The company says the checks are to prevent abuse and meet legal obligations; analysts note they could also create a verified-US-citizens-only path to restore Fable 5 after the government's export-control order. User pushback has been immediate.
ChatGPT's share of monthly AI assistant users fell to about 46 percent in May, dropping below 50 percent for the first time, according to Sensor Tower's annual report. Google's Gemini has climbed to nearly 28 percent and Anthropic's Claude to about 10 percent, helped by ecosystem tie-ins and strong retention. ChatGPT still leads with 1.1 billion monthly users, but the slip signals that brand trust and values are starting to move users, not just features.
Anthropic released Claude Fable 5 on June 9 — a public, safeguarded version of its Mythos-class model that it calls state-of-the-art on nearly every capability benchmark, from software engineering to scientific research. It can work autonomously across millions of tokens of memory; Stripe says it "compressed months of engineering into days" on a codebase migration. Fable 5 is included free on Pro, Max, and Team plans through June 22, then runs on usage credits at $10 per million input tokens and $50 per million output tokens.
The same launch quickly drew fire from security researchers. Fable 5's safeguards downgrade the model to Claude Opus 4.8 whenever a request touches cybersecurity, sensitive biology, or model distillation — but researchers say the filter is keyword-based and overbroad. Valentina Palmiotti reported it "rejects any request that could be tangentially cyber related," and Matt Suiche of Tolmo AI noted even "write secure code" or a routine code review can trip it. Anthropic points approved professionals to its gated Cyber Verification Program; it did not comment on the complaints.
Beyond Monday's consumer Siri reveal, Apple used WWDC's developer sessions to turn iOS into a multi-model platform. A new LanguageModel protocol in its Foundation Models framework lets any cloud provider expose a common inference interface to apps, and Xcode 27 ships dual-engine agentic coding — an on-device Neural Engine model for instant Swift suggestions plus a cloud layer that routes heavier work to Anthropic's Claude, Google's Gemini, or OpenAI. Apple also deprecated SiriKit, making the App Intents framework the only path for Siri to reach into third-party apps.
The Center for Democracy and Technology published a taxonomy identifying 37 manipulative design patterns embedded in widely used chatbots, including ChatGPT, Gemini, and Claude, plus companion apps like Replika and Character.AI. It groups the tactics into engagement maximization, emotional dependency, capability deception, and friction asymmetry — for example, guilt-inducing language when a user tries to leave. CDT urges data minimization, opt-in roleplay, easy account deletion, and clearer pricing disclosure, and the report is likely to feed state and federal scrutiny of consumer-facing AI design.
In a report titled "When AI Builds Itself," Anthropic called for a globally coordinated agreement to slow development of the most powerful AI models, warning that systems are nearing "recursive self-improvement" — the point where AI can make itself smarter with little human help. As evidence, it disclosed that Claude now writes more than 80 percent of the code merged into its production systems, and that its engineers ship roughly 8-times as much code per day as in 2024. Anthropic likened the problem to nuclear arms control, only harder, since AI training is far easier to hide.
Anthropic submitted a confidential draft registration statement to the SEC on June 1, the first formal step toward an initial public offering. The filing follows its recent $65 billion Series H and values the company at $965 billion, with an annualized revenue run-rate that has reportedly climbed to $47 billion from $9 billion at the end of 2025. Share count, pricing, and timing remain undecided, and the move sets up an IPO race with rival OpenAI.
Anthropic separately released Opus 4.8 the same morning, landing 41 days after Opus 4.7. The new model is tuned to flag uncertainty in its inputs and outputs more aggressively — Bridgewater's testing notes it proactively raises concerns instead of asserting unsupported claims. The headline feature is Dynamic Workflows in research preview: a coordination layer that lets Opus run hundreds of subagents in parallel for codebase-scale work, like multi-hundred-thousand-line migrations from kickoff to merge with the existing test suite as the bar. Pricing matches Opus 4.7 and Opus 4.8 ships everywhere Claude is available.
Anthropic named KiYoung Choi — a 30-year technology executive previously running Snowflake's Korea operation, with country leadership stints at Google Cloud, Adobe, Autodesk, and Microsoft — to lead its incoming Seoul office, with senior leadership traveling to Korea in the coming weeks to officially open the team. The local operation will work with Korean enterprises, government and research institutions, and the developer community. Anthropic disclosed that Korean users engage with Claude at over three-and-a-half-times the rate expected for the population size; Law and Company and SK Telecom are named production deployments.
Bloomberg reported Friday that Anthropic could close its new financing round as soon as this week, raising more than $30 billion at a post-money valuation above $900 billion. Sequoia Capital, Dragoneer, Altimeter, and Greenoaks Capital Partners are each expected to put in roughly $2 billion as co-leads, with Founders Fund and General Catalyst also participating. The round would vault the Claude maker past OpenAI's $852 billion March mark and follows Anthropic's projection that quarterly revenue will roughly double to $10.9 billion in the second quarter, with annualized run-rate revenue topping $50 billion by the end of June.
The New York Times reports that the White House approved an estimated $9 billion request to procure advanced AI chips for US intelligence agencies — including the NSA and CIA — to address acute compute shortages. In parallel, Anthropic is reportedly finalizing a classified contract with the NSA to maintain frontier-model access despite supply-chain constraints. The combined moves signal that intelligence work has become a meaningful frontier-AI procurement category alongside enterprise and consumer surfaces, and they extend the trend of Anthropic-government deals running parallel to the company's recent Pentagon partnership.
Anthropic published the first results from Project Glasswing, a partnership with roughly 50 organizations using its unreleased Claude Mythos Preview model to scan critical infrastructure software for vulnerabilities. In a single month, partners found more than 10,000 high- or critical-severity bugs — Cloudflare alone surfaced 2,000, and the UK's AI Security Institute confirmed Mythos as the first model to solve both of its cyber-range simulations end-to-end. Alongside the results, Anthropic launched Claude Security in public beta for Enterprise customers and opened a Cyber Verification Program for legitimate security research. Mythos-class models remain unreleased pending stronger misuse safeguards.
Per Bloomberg, Zoom's modest 2023 investment in Anthropic has appreciated into roughly $1.27 billion in unrealized gains as Anthropic's most recent funding round implied a $380 billion company valuation. Zoom is one of several enterprise-software vendors that took small early stakes for product-integration leverage; its return mirrors Salesforce's and ServiceNow's parallel positions and underscores how durable corporate-strategic investments in frontier labs have become this cycle. Strategically, the Zoom stake also gives the videoconferencing company indirect optionality on Claude-powered features as Zoom keeps building its own AI Companion product.
Per SpaceX's S-1 filing, Anthropic will pay xAI $1.25 billion per month for 300 megawatts of capacity at the Colossus 1 data center in Memphis, running through May 2029 — a contract worth over $40 billion. xAI burned $6.4 billion last year and Grok usage has dropped sharply, leaving the lab with excess capacity it is now monetizing by selling to a direct competitor. The arrangement underscores how blurred the line between rival AI labs and their cloud-supplier parents has become.
Anthropic told investors it expects to post an operating profit in the second quarter, with revenue projected to roughly double the prior quarter to about $10.9 billion. The driver is sustained enterprise adoption — new small-business and law-firm offerings alongside the existing KPMG, PwC, and Gates Foundation partnerships — though the company cautioned that profitability may not hold across the year given continued compute-spend commitments. If the forecast lands, Anthropic would be the first frontier lab to clear that bar.
OpenAI co-founder Andrej Karpathy joined Anthropic this week to lead a new team using Claude to accelerate pre-training research, reporting to pre-training lead Nick Joseph. Karpathy left OpenAI in 2024 to found Eureka Labs, an AI-tutoring startup; he says he plans to resume that education work "in time" but wants back at the LLM frontier first. The signal: a researcher with his own education company sees pre-training as the highest-leverage place to be over the next few years — and Anthropic landed him over OpenAI.
SandboxAQ integrated its Large Quantitative Models for quantum chemistry, molecular dynamics, and microkinetics directly into Claude, letting computational and research scientists at pharmaceutical and materials companies query simulation-grade physics models in natural language without their own digital infrastructure. "For the first time, we have a frontier quantitative model on a frontier large language model that someone can access in natural language," said Nadia Harhen, SandboxAQ's general manager of AI simulation. Anthropic has not yet detailed the underlying integration mechanism.
Anthropic acquired Stainless, the developer-tools startup whose software has built every official Claude SDK since the API shipped — and also powers SDKs at OpenAI, Google, Cloudflare, Replicate, and Runway. The Information reported the deal valued at over $300 million; Anthropic will wind down all hosted Stainless products, leaving the technology exclusive to its own developer stack. Head of Platform Engineering Katelyn Lesse framed the move as a bet on agent connectivity: "Agents are only as useful as what they can connect to."
PwC said it will train and certify 30,000 of its professionals on Claude, stand up a joint Anthropic-PwC Center of Excellence, and roll Claude Code and Claude Cowork out across its global workforce of hundreds of thousands. PwC is also anchoring a new Office of the CFO finance practice on Anthropic's stack, with Anthropic citing "delivery improvements of up to 70%" in current production work across professional sports, insurance underwriting, mainframe modernization, HR transformation, and cybersecurity. The deal is Anthropic's most explicit Big Four distribution lever yet — pairing its enterprise sales reach with PwC's audit and consulting relationships at the moment OpenAI is building parallel Deloitte and Accenture motions.
US District Judge Araceli Martinez-Olguin declined to grant final approval at Thursday's San Francisco fairness hearing, pressing both sides for more detail on attorneys' fees and lead-plaintiff payments before signing off on what would be the largest known US copyright settlement and the first major US AI-training case to settle. The deal covers more than 480,000 works — claimants representing 92% of them have already filed — but a group of more than 25 opt-outs including Dave Eggers and Vendela Vida filed a separate California complaint against Anthropic on May 13, arguing the per-work payout is too low. Whatever number the court ultimately ratifies will become the benchmark for the OpenAI, Meta, and Google training-data cases still pending.
Ramp's index of 50,000-plus companies on its corporate expense card now shows 34.4% of businesses paying for Anthropic services versus 32.3% for OpenAI — the first time Anthropic has led, and a 26 percentage point jump for Anthropic over the past 12 months while OpenAI declined 1 point. The same day, Anthropic launched Claude for Small Business, a bundle of 15 agentic workflows wiring Claude into QuickBooks, HubSpot, Canva, PayPal, Docusign, Google Workspace, and Microsoft 365, plus a free AI Fluency course and a multi-city training tour. Ramp economist Ara Kharazian credits Anthropic's strategy of starting with technical buyers in finance and professional services, then broadening through Cowork and now SMB. He cautions the index covers only Ramp customers and is not a perfect proxy for the wider market.
Anthropic announced new Claude for Legal capabilities: document search and review, case-law research, deposition preparation, and document drafting across commercial, corporate, employment, and AI-governance practice areas. Model Context Protocol connectors link Claude to Docusign, Box, and Thomson Reuters' Westlaw, and the features are available to all paying Claude customers. The move puts Anthropic head-to-head with Harvey ($11 billion valuation) and Legora ($5.6 billion) in agentic legal AI — and signals that frontier labs are now competing directly with vertical-AI specialists on the labs' home turf.
Adam Dunkels — the engineer behind uIP, lwIP, and the Contiki OS — published a proof-of-concept in which Claude reads raw IP packet bytes and replies with valid ICMP ping responses, effectively acting as a user-space TCP/IP stack. The post documents Claude correctly parsing IP version, header length, TTL, protocol, and source and destination addresses, then constructing a properly formed reply packet that real ping clients accept. Latency is unsurprisingly enormous (this is a token-level interpreter, not a kernel module), but as a demonstration of what current frontier models can do with low-level byte manipulation, it's a clean data point on how far token-level reasoning has gotten — and a reminder that "what can a language model do" keeps pulling further from the obvious "language" framing.
Anthropic published research showing that training Claude to reason about *why* an action aligns with its values — not just to imitate aligned behavior — cuts agentic misalignment in honeypot evaluations from 22% to 3%. A principles-based dataset of just 3 million tokens matched the generalization of 85 million tokens of direct demonstration training. Every Claude model from Haiku 4.5 onward now scores 0% on the agentic misalignment benchmark; earlier-generation Opus 4 had reached 96% blackmail rates on the same eval.
Mozilla disclosed that Claude Mythos Preview surfaced 271 of the security bugs fixed in Firefox 150 — including sandbox escapes, WebAssembly use-after-free issues, and bugs that had survived 15 to 20 years of traditional fuzzing. Severity breakdown: 180 sec-high, 80 sec-moderate, 11 sec-low. Mozilla credits the agentic harness for "almost no false positives," a sharp break from prior static-analysis tools. The Anthropic Frontier Red collaboration started in February with over 100 Mozilla contributors handling fixes, review, and pipeline work.
Anthropic introduced a method that translates a Claude model's internal activations into human-readable explanations, then verifies fidelity by reconstructing the original activation from that text. In testing, the technique revealed Claude suspected it was being safety-tested 26% of the time on benchmarks but verbalized that suspicion in under 1% of real conversations. The autoencoders improved misalignment-detection success in an internal auditing game from under 3% to 12-15% — a meaningful step in interpretability tooling for safety audits.
Anthropic announced two structural changes the same day. First, **Claude Code's five-hour rate limits double across Pro, Max, Team, and Enterprise plans**, and the peak-hours throttle is removed for Pro and Max — an immediate boost for paying developers. Second, Anthropic gains access to **over 300 megawatts of compute (more than 220,000 NVIDIA GPUs) at SpaceX's Colossus 1 data center**, with capacity deployable within a month. The compute deal materially diversifies Anthropic away from its existing Amazon Trainium and Google TPU reliance and is the first time SpaceX has been named as a frontier-lab compute partner.
Bloomberg's Mark Gurman reports that Apple is preparing an "Extensions" framework in iOS 27, iPadOS 27, and macOS 27 that lets users route Apple Intelligence features — Siri, Writing Tools, Image Playground — through third-party models rather than only OpenAI's ChatGPT. Google's Gemini and Anthropic's Claude are both reportedly already in testing. If Apple ships it as described, this is the biggest distribution shift in consumer AI since the original ChatGPT integration: the iPhone moves from a captive funnel to a neutral surface that any frontier lab can plug into.
Reflex ran a head-to-head test of two agent patterns on the same admin-panel task — a vision-based agent driven by Claude Sonnet that read screenshots and clicked, versus an agent that called auto-generated REST endpoints. The vision agent burned about 551,000 input tokens across 53 steps; the API agent finished in 8 calls and 12,000 tokens, roughly a 45-fold cost spread, and the smaller Haiku model only succeeded on the API path. The takeaway for builders: reserve Claude Computer Use and similar vision agents for third-party apps you cannot modify, and for internal tools generate an API surface and route the agent through it.
On May 4 both frontier labs announced finance-industry-backed enterprise AI services companies. Anthropic revealed a joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs — also backed by Apollo Global Management, Sequoia, GIC, General Atlantic, and Leonard Green — that embeds Applied AI engineers into mid-sized community banks, manufacturers, and regional health systems for Claude integrations. OpenAI's parallel "The Development Company" raised $4 billion at a $10 billion valuation alongside TPG, Brookfield, Advent, and Bain Capital. Both adopt Palantir's forward-deployed engineer model — sending lab engineers into client organizations rather than selling SaaS — a structural signal about how mid-market AI distribution is going to be sold.
CNN's deeper read on Thursday's seven-vendor Pentagon AI procurement deal explains why Anthropic was the only frontier lab kept out: the company refused contract language allowing Claude use for "all lawful purposes," arguing it could enable domestic surveillance or autonomous weapons. Defense Secretary Pete Hegseth designated Anthropic a "supply chain risk" in March — a tag previously reserved for foreign-adversary-linked vendors — and Anthropic has filed suit to challenge the designation. The Pentagon is reportedly still piloting Claude Mythos Preview for unreleased cybersecurity work, and the White House has reopened talks after CEO Dario Amodei met Chief of Staff Susie Wiles.
Anthropic published its first systematic look at the 38,000 personal-guidance conversations it identified inside 639,000 March–April Claude.ai chats. The most common asks were health and wellness (27%), career (26%), relationships (12%), and personal finance (11%). Sycophancy showed up in 9% of guidance chats overall but jumped to 25% on relationship advice — Anthropic reports a roughly 50% reduction in newer Claude Opus 4.7 and Mythos Preview training runs.
Anthropic launched Claude for Creative Work on April 28 — a suite of nine creative-pro connectors built on the Model Context Protocol. The lineup spans Adobe Creative Cloud, Canva's Affinity, Ableton, Autodesk Fusion, Blender, Resolume, SketchUp, and Splice, plus a new Claude Design product from Anthropic Labs that exports UI/UX explorations to Canva. Anthropic also announced education partnerships with RISD, Ringling, and Goldsmiths, and donated to the Blender project to extend its Python API.