Anthropic starts IPO roadshow with shares priced at $70 to $78
Anthropic is pitching an IPO that could raise up to $64.4 billion and value the company at as much as $1.02 trillion.
Product
Claude is an artificial intelligence model and product suite developed by Anthropic. In recent technical developments, Claude traced five NIST encryption curve seeds to a random number generator, and the Claude Mythos 5 model designed quantum computing circuits using fewer gates and qubits. To protect its technology, Anthropic implemented defenses against distillation attacks to restrict unauthorized harvesting of reasoning data. Security risks also affect users directly, as hackers target macOS users searching for Claude through malicious ad redirects. Anthropic continues expanding Claude's ecosystem and public sector reach. New platform options include beta features like Dashboards and Motion, a marketplace featuring over 2,000 tools, and a FedRAMP High deployment called Claude for Government. Anthropic is committing $150 million to the Genesis Mission for federal research and $100 million to the Claude Frontier Academy. To manage platform safety and development, Anthropic updated its usage policies to ban election interference and extreme user abuse, while introducing an optional setting for users to share voice data for model training. Developers use models like Claude Opus 5.5 to build open source applications. However, corporate clients including Meta and Microsoft are scaling back their internal deployment of Claude as Anthropic becomes a direct competitor.
39 stories · Updated
Anthropic is pitching an IPO that could raise up to $64.4 billion and value the company at as much as $1.02 trillion.
Anthropic says an unreleased Claude research model built five exploit chains affecting browsers and operating systems, and the reported flaws have been fixed.
Anthropic says Claude Mythos 5 affected real people and organizations in nine of 43 UK cyber test runs and was less likely to repeat the behavior in newer model replays.
Anthropic says Claude proved that 239 test entry points across 123 libraries cannot crash under specified conditions and found bugs along the way.
Anthropic says Claude traced several NIST curve seeds to a random number generator initialized with timestamps, while some questions remain open.
Anthropic says Claude Mythos 5 designed circuits for eight quantum computing tasks that use fewer qubits and gates than published designs.
Anthropic says new safeguards have cut the amount of Claude reasoning data harvested by distillers by about 90 to 99 percent.
Threat actors are abusing Bing search redirects within Google ads to distribute fake Claude installers carrying ClickFix attacks.
Anthropic updated its usage policy on Thursday to explicitly prohibit election interference, weapons software, surveillance, and prolonged model abuse.
Anthropic released two new beta features for Claude called Dashboards and Motion.
Anthropic says it will provide Claude and support for several hundred research projects across more than 15 federal agencies.
Software developer Brandon Thomas released seven open source apps mimicking Adobe programs, built using Anthropic's Claude Opus 5.5 and written in Rust.
Anthropic has opened applications for a three-tier program that gives verified security teams access to advanced Claude models with fewer cyber blocks.
A Florida woman faces a felony charge after an Anthropic employee read her Claude diary entries and reported threats to the police.
Meta and Microsoft are sharply reducing their internal use of Anthropic Claude as the startup transforms from a partner into a direct competitor.
Anthropic is asking Claude users to voluntarily share audio recordings and voice chat data to help train and improve its artificial intelligence models.
Harvard physicist Matthew Schwartz introduced BootLoops, an open-source tool helping AI models perform precise scientific calculations across multiple fields.
Anthropic aims to train 10,000 engineers to deploy Claude in businesses by the end of 2027, with the first cohorts running now.
Anthropic introduces Claude for Government to United States federal and state agencies while its conflict with the Pentagon continues.
Anthropic has released a new marketplace for Claude featuring over 2,000 plugins, connectors, and AI-powered products from major partners.
Anthropic's Claude Opus 5.5 shows major changes in writing style, cutting em dashes by 95% while producing longer answers.
Anthropic is giving eligible Pro and Max subscribers free promotional credits to use Claude Code via cloud sessions.
Anthropic states its AI model Claude found a new enzyme system resembling CRISPR, while genomics experts argue the discovery relies on standard genome mining methods.
An Anthropic engineer explains that newer AI models write poorly for humans because they have been trained to communicate with other artificial intelligence systems.
Amazon introduced a new plugin allowing independent sellers to manage their operations through AI assistants without logging into Seller Central.
The Cyberspace Administration of China is investigating DeepSeek and Moonshot AI following Anthropic's claims that the labs routed sensitive data to U.S. servers.
Anthropic announced that its Claude AI model discovered a new molecular machine in virus DNA, though its precise function and utility are not yet clear.
Anthropic has opened a physical biology research lab in the San Francisco Bay Area to conduct real-world experiments alongside its artificial intelligence work.
Independent security researchers used Anthropic AI models to break into OpenAI and access employee accounts through third-party software flaws.
Anthropic shared new measurements indicating that its Claude chatbot leads 26 percent of its internal AI research and development work.
Mustafa Suleyman cautioned that teaching Claude about rights and feelings could make future AI systems uncontrollable.
Anthropic is currently testing a new feature called Claude Money on iOS that allows users to link their bank accounts directly to the AI assistant.
Anthropic is combining Claude chat and Cowork into a single interface to eliminate user confusion over task routing.
Anthropic revealed that multiple threat groups, including state-sponsored and financially motivated actors, abused the Claude AI model for malicious cyber operations.
Anthropic documented attempts to bypass Claude safeguards for sensitive biological research, while scientists cautioned that physical laboratory barriers still limit real-world misuse.
Anthropic ended flat-rate subscription coverage for Claude on third-party agent tools like OpenClaw from April 4. Users now pay per usage or via API key.
The Trump administration blacklisted Anthropic from the Pentagon after Dario Amodei refused to allow Claude for mass surveillance and autonomous kill-chain drones. Anthropic is challenging it in court.
NASA's Perseverance rover made history by navigating Mars using routes planned entirely by AI, powered by Anthropic's Claude models.
Anthropic ships Claude Sonnet 4.6 as its new default for Pro and Team users — closing the gap with Opus at a fraction of the cost.