Skip to content

As of 5 October 2026.

This dossier collects Beckmann's reporting on Anthropic. Each entry is one sentence from the original report. Where that report stays indexed, the link points to it. The other reports are summarised here.

Timeline

  • 5 July 2026. Anthropic launches its own preclinical drug discovery for neglected diseases and presents Claude Science.
  • 10 July 2026. Anthropic appoints former Fed Chair and Nobel laureate Ben Bernanke to its Long-Term Benefit Trust – background, criticism, and context.
  • 10 July 2026. Anthropic Invites AI Questions After 52,000-Person Survey. Anthropic surveyed 52,000 Americans and 81,000 Claude users on AI fears, then launched 'Hard Questions' to collect public concerns.
  • 17 July 2026. Morgan Stanley, Goldman Sachs and JPMorgan are accompanying the process – the valuation is at 965 billion dollars.
  • 8 August 2026. Claude Code activates auto mode by default starting August 14. A safety classifier will take over command verification in Claude Code, and according to Anthropic, it recognizes significantly more risks than human oversight.
  • 16 August 2026. Banks are evaluating the planned IPO in October based on a forecast that is expected to more than quadruple by 2028.
  • 16 August 2026. A test pitting rival Claude agents against each other ends in account lockouts and disguised code, with only newer models de-escalating.
  • 21 August 2026. Reuters reports the AI provider is easing centralized data storage rules for corporate customers, after criticism from regulated industries.
  • 25 August 2026. Six investors expect a record valuation for the October IPO, backed by rapidly rising revenue figures.
  • 26 August 2026. The Wall Street Journal reports: the prospectus draft exceeds SpaceX's record estimate – analysts consider the figure unrealistic.
  • 26 August 2026. The AI assistant automatically remembers project details across chat and the task tool Cowork from now on – adjustable in the topic overview.
  • 28 August 2026. A US federal judge overturns the security risk classification of Anthropic by the Pentagon – the contract exclusion remains in place for now.
  • 28 August 2026. Anthropic opens standard for AI-controlled laboratory robots. Anthropic is testing a common interface for AI agents and physical machines with Genentech, QuEra, and other partners.
  • 29 August 2026. First details on the largest AI IPO: Bankers calculate up to two trillion dollars, early investors could sell for the first time.
  • 30 August 2026. Claude Code: Researcher bypasses auto mode with malicious code. A security researcher bypasses Claude Code's automatic release function using a tampered ZIP file – with up to 80 percent success rate in tests.
  • 2 September 2026. The US Department of Defense opens its AI portal for ChatGPT Mil and Grok for Government – Anthropic's Claude remains excluded despite a court ruling.
  • 3 September 2026. Corporate clients will manage their usage data in their own cloud in the future, while Anthropic continues to automatically search for abuse.
  • 7 September 2026. The prospectus for the largest AI IPO of the year is coming later than planned – new details on schedule, valuation, and credit line.
  • 12 September 2026. The chip manufacturer is reportedly set to become an anchor investor in what could be the largest IPO in history.
  • 19 September 2026. The AI company operates a bio-lab for physical experiments, according to Reuters – among the pharmaceutical partners are Novo Nordisk and Genentech.
  • 20 September 2026. The AI company's revenue has grown by half in two months – new figures fuel the IPO planned for November.
  • 20 September 2026. Anthropic launches Claude Docs and Claude Slides in chat. Anthropic ends Cowork as a standalone tool and brings document and slide creation directly into the Claude chat.
  • 25 September 2026. A new class of shares is intended to permanently secure the majority of votes for the seven company founders – with minimal capital share.
  • 26 September 2026. Anthropic discovers a new enzyme system in the genome with Claude. Almost 1,000 autonomous Claude agents found an unknown enzyme system with CRISPR-like repeats in bacterial viruses.

What this means for companies in Germany

Teams introducing Claude should follow the reports that remain indexed. Price, availability and contract details stay in those articles, not in this overview. Entries without a link replace the separate report.

All stories