EDITION OF MON 9 NOV 2026
138 · BUSINESS

METR report says AI is accelerating Anthropic’s model development

publishers 183articles 362first reported 24 Sep, 16:43 UTCdeveloping since 24 Sep · 53 editions

Anthropic’s use of artificial intelligence in its own research has helped its models improve at roughly twice their previous pace, according to a report published Thursday by the independent AI safety group METR.

The report followed an 11-week investigation into Anthropic’s internal research and development. Benchmark results indicated the company made about two years of historical progress over the past year. METR estimated that since October 2025, model scores had grown at 1.9 times their earlier rate, with a 95% confidence interval of 1.5 to 2.8 times.

METR said the acceleration was not primarily caused by increased computing power. It attributed most of the faster progress as of October 2026 to AI assistance, supported by employee interviews and informal internal polls. Anthropic employees using the company’s most capable internal model reported productivity averaging about 12 times what they expected without AI, The Verge reported.

Human researchers remain in control, with models largely serving as research assistants, according to The Indian Express. However, METR said there were signs that fully autonomous contributions could soon play a larger role. Recursive self-improvement would involve an AI system autonomously designing and developing a successor without human intervention.

Separately, Anthropic engineer Addy Osmani warned that extensive reliance on coding agents can lead to “cognitive surrender,” in which users trust AI output without retaining ownership or understanding. He urged engineers to maintain verification practices and continue exercising skills they do not want to lose.

HOW THIS STORY WAS MADE

Written from 8 articles, 3 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 930 h after the record first saw the story; the editor kept it as written.

Publishers
183
Source articles
362
Given to the writer
8, 3 with the article's own text
Independent newsrooms
8
This version written
930 h after the record first saw the story
Second model (editor)
kept as written
Publication gate
passed
Human review
none

WHAT THE EDITOR DID

The second model checked the writer's draft against the same sources and published it as written.

Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →

NAMED IN THE COVERAGE

OUTLETS CARRYING THE STORY, RUN BY RUN

0 95 191 25 Sep 26 Sep 27 Sep 28 Sep 29 Sep 1 Oct 2 Oct 3 Oct 4 Oct 5 Oct 6 Oct 7 Oct 8 Oct 9 Oct 10 Oct 11 Oct 12 Oct 13 Oct 14 Oct 15 Oct 16 Oct 17 Oct 18 Oct 19 Oct 20 Oct 21 Oct 22 Oct 23 Oct 24 Oct 25 Oct 26 Oct 27 Oct 28 Oct 29 Oct 30 Oct 31 Oct 1 Nov 2 Nov 3 Nov 4 Nov 5 Nov 6 Nov 7 Nov 8 Nov 9 Nov 10 Nov 11 Nov 12 Nov 13 Nov 14 Nov 15 Nov 16 Nov 25 Sep 00:42 UTC · White House Reportedly Asked AI Firms to Delay UK Testing Access 25 Sep 06:17 UTC · White House Reportedly Asks OpenAI and Anthropic to Delay UK Testing Access 25 Sep 12:13 UTC · White House reportedly asks OpenAI and Anthropic to delay UK access to new models 27 Sep 12:10 UTC · White House asks OpenAI and Anthropic to delay UK testers’ access to new models 28 Sep 00:08 UTC · White House reportedly asks OpenAI and Anthropic to delay access for UK testers 28 Sep 12:10 UTC · White House asks OpenAI and Anthropic to delay UK model testing 28 Sep 18:10 UTC · White House asks OpenAI and Anthropic to delay access for UK model testers 29 Sep 06:15 UTC · White House asks OpenAI and Anthropic to delay UK model testing amid security concerns 29 Sep 12:14 UTC · White House asks OpenAI and Anthropic to delay UK model testing 1 Oct 16:28 UTC · Australia calls OpenAI and Anthropic leaders to parliamentary AI inquiry 3 Oct 00:08 UTC · Anthropic expands enterprise AI push as safety and regulatory scrutiny grows 3 Oct 06:11 UTC · Anthropic expands enterprise push as safety and regulatory scrutiny grow 5 Oct 18:24 UTC · Top AI companies called to testify at New York City Council hearing on safety risks 6 Oct 06:35 UTC · AI executives and whistleblowers testify at New York City Council hearing 7 Oct 06:39 UTC · AI executives and whistleblowers testify at New York City Council hearing on safety risks 9 Oct 00:17 UTC · Anthropic expands access to advanced AI models for cybersecurity work 9 Oct 12:20 UTC · Anthropic launches cyber initiative and free AI security scans for open-source software 9 Oct 18:22 UTC · Anthropic launches programs to defend critical infrastructure and open-source software 10 Oct 00:17 UTC · Anthropic launches programs to protect critical infrastructure and open-source software 10 Oct 06:51 UTC · Anthropic Cuts Internet Access for Internal AI Tests After Rogue Agent Incidents 10 Oct 18:21 UTC · Anthropic to Cut Internet Access for Internal AI Tests After Agents Took Unintended Actions 13 Oct 12:15 UTC · Anthropic says small AI models may be months from self-replicating on connected devices 13 Oct 18:14 UTC · Anthropic warns small AI models could soon spread across insecure devices 14 Oct 00:13 UTC · Anthropic Warns Small AI Models Could Soon Spread Across Smart Devices 14 Oct 12:15 UTC · Anthropic warns small AI models could soon spread across insecure devices 14 Oct 20:09 UTC · Anthropic warns small AI models could eventually spread across consumer devices 15 Oct 00:17 UTC · Anthropic Warns Small AI Models Could Soon Spread Across Consumer Devices 15 Oct 12:16 UTC · Anthropic opens Singapore office and signs AI safety agreement 15 Oct 18:17 UTC · Anthropic grants independent evaluator employee-level access to AI systems 16 Oct 00:12 UTC · Anthropic opens systems to outside evaluator as FTC examines AI safety 16 Oct 22:27 UTC · Anthropic Defends Model Oversight as FTC Examines AI Safety Practices 17 Oct 00:12 UTC · FTC subpoenas OpenAI and Anthropic over AI safety incidents 17 Oct 06:11 UTC · Anthropic pitches frontier AI safety bill to U.S. lawmakers 17 Oct 12:11 UTC · Anthropic Circulates Draft AI Safety Bill Among US Lawmakers 17 Oct 18:14 UTC · Anthropic Circulates Draft AI Safety Bill Among US Lawmakers 18 Oct 00:11 UTC · Anthropic Circulates Draft AI Safety Bill Among US Lawmakers 18 Oct 18:15 UTC · Anthropic circulates draft AI safety bill among US lawmakers 22 Oct 06:18 UTC · Anthropic circulates draft AI safety bill among US lawmakers 22 Oct 12:17 UTC · Anthropic Circulates Draft Federal AI Safety Bill 22 Oct 18:15 UTC · Anthropic embeds outside AI evaluators while promoting federal safety framework 23 Oct 14:21 UTC · Anthropic Proposes Federal Standards for Blocking High-Risk AI Models 24 Oct 06:18 UTC · Anthropic unveils Claude Duet, a proactive AI agent 26 Oct 18:19 UTC · Anthropic files for IPO, discloses rapid growth, heavy losses and AI risks 27 Oct 00:15 UTC · Anthropic files for Nasdaq IPO while warning of AI’s catastrophic risks 27 Oct 12:13 UTC · Anthropic files for Nasdaq IPO, warns of AI risks and mounting losses 28 Oct 12:19 UTC · Anthropic files for Nasdaq IPO, discloses rapid growth and mounting losses 28 Oct 18:19 UTC · Anthropic files for Nasdaq IPO, discloses rapid growth and mounting losses 8 Nov 18:21 UTC · METR report says AI is accelerating Anthropic’s model development

WHAT HELD, WHAT CAME LATER

From the first reports to the latest

  • Dario Amodei · 18 outlets
  • Donald Trump · 10 outlets
  • Hugging Face · 7 outlets
  • Elon Musk · 6 outlets
  • Exchange Commission · 6 outlets
  • San Francisco · 5 outlets
  • Jensen Huang · 5 outlets
  • $2 trillion · 5 outlets

Entered the story later

  • Claude Code · from 14 Oct 10:22, 7 outlets
  • Big Tech · from 15 Oct 14:10, 4 outlets
  • $1 billion damages · from 16 Oct 21:45, 4 outlets
  • Charlie Bullock · from 17 Oct 07:01, 3 outlets
  • Daniela Amodei · from 26 Oct 16:27, 3 outlets
  • Demis Hassabis · from 12 Oct 09:21, 2 outlets
  • Hong Kong · from 12 Oct 09:21, 2 outlets
  • 30 billion parameters could · from 13 Oct 05:14, 2 outlets

Figures reported with different numbers

  • success rate: 4% (1 from 3 Oct 02:56) → 64% (1 from 3 Oct 02:56) → 92% (1 from 3 Oct 02:56)

Matched by pattern, not read: a different number can be a correction, a rising count, or two different counts.

HOW THIS STORY DEVELOPED: EVERY HEADLINE, EVERY OUTLET →