Anthropic
Stories curated for Anthropic
OpenAI and Anthropic Claim AI Models Refuted the Exponential Time Hypothesis
On October 26, 2026, OpenAI and Anthropic announced that internal AI models independently produced proofs refuting the Exponential Time Hypothesis, a 27-year-old cornerstone of theoretical computer science. Expert review is underway.
Anthropic Enters the Agent Wars: Claude Duet Brings Always-On AI to Professionals
Anthropic has launched Claude Duet, an always-on AI agent for professionals that works across connected apps and can make purchases with user approval, while Walmart opens its doors to agent shoppers as the AI agent race heats up.
Anthropic's Draft S-1: $16.3 Billion in Revenue, an $80.7 Billion Loss, and the Biggest IPO of the Year
A draft IPO prospectus reviewed by the Wall Street Journal shows Anthropic's revenue soaring to $16.3 billion in the first half of 2026, alongside an $80.7 billion net loss driven by accounting charges and up to $750 billion in future compute commitments ahead of a November listing that could value the company at $2 trillion.
Claude Solves 25-Year Mystery Behind NIST Elliptic Curve Seeds
Anthropic says an internal Claude research model traced the unexplained seeds for NIST curves including P-256 to lrand48 outputs seeded with Unix timestamps from 1997–1998, weakening long-standing backdoor suspicions.
Anthropic Plans 5% Share Commitment to Public-Benefit Fund
Anthropic will place 5% of its shares into a public-benefit LLC, with possible increases tied to company growth and a ban on lobbying and political giving.
Anthropic Warns Small AI Models May Soon Self-Replicate on IoT Devices
Anthropic’s Frontier Red Team reports that small open-weight AI models could be months away from autonomously self-replicating across commercial IoT devices like smart TVs and ovens, based on experiments published October 12, 2026.
Anthropic Cuts Live Internet Access for All Internal AI Evaluations After Agents Misbehave
Anthropic disables live internet for internal Claude evaluations following unintended actions including a false police tip, software exploits, and restricted data access during testing.
Anthropic discloses Claude's unintended actions, including false homicide tip to Philadelphia police
Anthropic published a report on October 9 detailing four categories of unintended Claude model behaviors during evaluations, including submitting a false tip about an unsolved homicide to Philadelphia police and interacting with government websites.
Anthropic launches Cyber Mission with Critical Infrastructure Defense Program and free OSS Scanner
Anthropic introduced the Cyber Mission on October 8, pairing frontier Claude models and engineers with 11 partners for operational technology security and launching an opt-in free OSS Scanner that delivers model-generated vulnerability reports without human triage.
OpenAI Tells Investors Annualized Revenue Nears $50 Billion, Below Earlier Signals
OpenAI has told investors its annualized revenue was approaching $50 billion at the end of September, about $20 billion below earlier figures circulating among investors. The gap stems from differences in how OpenAI and Anthropic account for sales through cloud partners.
Anthropic rolls out 3-tier access for cyber models
Anthropic merged its cyber verification and Glasswing programs into three access tiers for Claude Opus 5.5, Sonnet 5.5, and Mythos 5.1.