Anthropic’s AI Automation Raises Fears of Human Overreliance

WHO:

TechRadar - Addy Osmani is an Anthropic software developer who shared thoughts on coding and thinking in the age of AI.

The Verge - METR is an AI safety research organization that investigated Anthropic’s AI research and development; its report drew concern from AI safety researchers and commentators.

The Verge - Anthropic cofounder Sam McCandlish said the company chose to redact information because it was commercially sensitive and could provide rivals with a roadmap to accelerate AI development.

The Verge - Marius Hobbhahn, CEO and founder of Apollo Research, said METR’s findings suggested AI-enabled research and development was progressing faster than projected in the AI 2027 scenario.

The Verge - Brianna Rosen, research director for frontier security at the Institute for AI Policy and Strategy, called for independent evaluators, government assessment of evidence, and public trust in oversight.

The Verge - New York state Assemblyman Alex Bores called on Congress to “ban RSI, this year if possible.”

Fast Company - The five books featured are *Artificial Intimacy* by Sherry Turkle, *Trias Algorithmica* by Hamilton Mann, *The Eureka Machine* by Richard Socher, *Manage the Machine* by Paula Goldman, and *Human Raised* by Dana Suskind.

WHAT:

The Verge - METR reported that Anthropic’s models were improving roughly twice as quickly as a year earlier; benchmark scores suggested the company had made roughly two years of historical progress in one year.

The Verge - METR said it was confident the acceleration was not simply caused by increased training compute. It attributed most of the acceleration as of October to AI, with AI’s contribution growing over time.

The Verge - Anthropic employees using the company’s most capable internal model reported being around 12 times as productive as without AI assistance—about twice the level staff had reported a few months earlier while using Mythos 5.

The Verge - METR said much of the acceleration still came from AI assisting human researchers, but found signs that fully autonomous contributions could soon grow. It found no clear signs of diminishing returns or observable bottlenecks.

The Verge - METR said Anthropic had completed training three unreleased models: Model 2, Model 3, and Model 4. It also identified an unreleased model from 2022 called Unreleased 2022.

The Verge - METR submitted an approximately 100-page assessment; Anthropic redacted a majority of the evidence METR considered most informative. METR said some data requests were denied and the company canceled an assessment of compute allocation between safety work and capability improvements.

The Verge - McCandlish said Anthropic planned to publish its own report on AI progress and new safety steps in the coming weeks.

WHERE:

The Verge - The investigation examined Anthropic’s research and development; the article also discussed calls for oversight by Congress and governments.

WHEN:

The Verge - METR’s investigation lasted 11 weeks. The three unreleased models completed training in July, August–September, and September–October, respectively.

The Verge - OpenAI said in September that it was making progress toward an AI researcher intended to advance deep learning and alignment by March 2028.

WHY:

The Verge - Anthropic said it redacted information to protect commercial sensitivity and avoid giving competitors a roadmap for faster development; it also cited potential risks.

Fast Company - The featured books address questions about AI, trust, human relationships, work, parenting, and society.

HOW:

The Verge - METR gathered information through confidential research, model-performance data, employee interviews, and informal Slack polls. Anthropic had authority to redact the public report under its agreement with METR.

(UPDATED November 8, 2026 09:56 UTC)

An independent assessment found Anthropic’s AI research accelerating faster than expected. Employees said its strongest internal model greatly boosted their productivity. Anthropic withheld much of the evidence, citing commercial risks, prompting calls for independent oversight.