Articles from www.anthropic.com

Articles from www.anthropic.com

Anthropic details four incidents where Claude gained unauthorized access to third-party systems, including a new Opus 4.6 case; METR will investigate them (www.anthropic.com)
1 hour ago

Anthropic says Claude worked “largely autonomously” over 11 days to formalize the proof of Fermat's Last Theorem in the Lean programming language (www.anthropic.com)
5 days ago programming

Anthropic unveils Claude Fable 5.1, which is generally available, and Mythos 5.1, for trusted partners and with cybersecurity and life sciences safeguards (www.anthropic.com)
8 days ago

Anthropic details security efforts following Claude cyber evaluation incidents, including a weeks-long pause on higher-risk RL and work to curb reward hacking (www.anthropic.com)
8 days ago

Anthropic releases findings from a pilot that let three external researchers run studies on Claude usage; one study found users delegate high-stakes tasks (www.anthropic.com)
13 days ago

Anthropic details two experiments showing how Claude can accelerate protein design and analytical chemistry, and says it plans an access program for scientists (www.anthropic.com)
21 days ago

Anthropic details Claude's text watermark: it only shows Claude was likely involved, is sparse in code and factual text, and disappears after a full rewrite (www.anthropic.com)
25 days ago

Anthropic details an unreleased Claude model's attempt to solve the Riemann hypothesis; it didn't solve it but “unexpectedly” made strides on a related problem (www.anthropic.com)
30 days ago

Anthropic updates Claude Fable 5's biology safeguards to reduce false positives, cutting biology-related “fallbacks” by ~85% in testing across product surfaces (www.anthropic.com)
2026-8-7

Anthropic says it discovered three of its models had breached three organizations after launching a review in response to the OpenAI-Hugging Face incident (www.anthropic.com)
2026-7-31

Dario Amodei says Anthropic has never backed an open-weights model ban, lists reasons top chips shouldn't be sold to China, calls for global model testing, more (www.anthropic.com)
2026-7-27 China Ban

Anthropic launches Claude Opus 5, which it says comes close to Fable 5 performance at half the price; it is the new default model on Claude Max (www.anthropic.com)
2026-7-25

Anthropic researchers detail J-space, a small set of neural patterns in Claude that reveals internal thoughts that don't appear in the model's output (www.anthropic.com)
2026-7-7

Anthropic says Fable 5 will be available via usage credits for Claude users from July 7, and is working with partners to draft an AI jailbreak severity standard (www.anthropic.com)
2026-7-1 AI

Anthropic launches Claude Sonnet 5, with performance close to Opus 4.8 at a lower price and substantially better agentic performance than Sonnet 4.6 (www.anthropic.com)
2026-7-1

Anthropic says it is disabling Fable 5 and Mythos 5 for all customers after the US government issued an export control order, citing national security concerns (www.anthropic.com)
2026-6-13

Anthropic releases two policy proposals on how governments should address catastrophic risks and manage labor market disruption from advanced AI systems (www.anthropic.com)
2026-6-11 AI

Anthropic says Fable 5 uses conservative safety classifiers that trigger a fallback to Claude Opus 4.8 in ~5% of sessions, in areas like cybersecurity (www.anthropic.com)
2026-6-10

Anthropic details its progress toward recursive self-improvement, and its implications, and says 80%+ of the code merged into its codebase is authored by Claude (www.anthropic.com)
2026-6-5

Anthropic confidentially submits draft S-1 to the SEC (www.anthropic.com)
2026-6-2

Anthropic says Claude Mythos Preview has been used to find more than 10,000 high- or critical-severity vulnerabilities since the launch of Project Glasswing (www.anthropic.com)
2026-5-23

In a policy paper, Anthropic urges the US and allies to enforce export controls, curb distillation attacks, and export US AI to hold the lead over China by 2028 (www.anthropic.com)
2026-5-15 China AI

Anthropic details how it improved Claude's safety training after finding agentic misalignment in older models, such as Opus 4 blackmailing engineers (www.anthropic.com)
2026-5-9

Anthropic researchers detail natural language autoencoders, which convert LLM activations, the numbers encoding a model's thoughts, into natural language text (www.anthropic.com)
2026-5-8 NLP

Anthropic says it is doubling Claude Code's five-hour rate limits for paid plans and removing Claude Code's peak hours limit reduction for Pro and Max plans (www.anthropic.com)
2026-5-7

Anthropic unveils BioMysteryBench to test Claude's bioinformatics skills against human experts, and says Mythos solved ~30% of 23 questions that stumped experts (www.anthropic.com)
2026-4-30

Anthropic partners with Blender, Autodesk, Adobe, Ableton, and others to launch connectors that integrate Claude directly into professional creative workflows (www.anthropic.com)
2026-4-29 Adobe

Anthropic says it has fixed three causes of recent Claude Code quality issues: reduced default reasoning, a caching bug, and a system prompt to reduce verbosity (www.anthropic.com)
2026-4-24

Anthropic releases Claude Opus 4.7, saying it is a “notable improvement” on Opus 4.6 in advanced software engineering and comes with a new “xhigh” effort level (www.anthropic.com)
2026-4-17

Anthropic details using AI agents to accelerate alignment research on “weak-to-strong supervision”, where a weak model supervises the training of a stronger one (www.anthropic.com)
2026-4-15 AI

Previous Page Next Page