Anthropic– tag –

Anthropic and its partner Theorem used Claude to formally verify 123 open-source libraries, generating about 180 million lines of proof code. We explore what it means for AI to move from “writing” to “proving”—and where its limits lie.

Claude’s official account announced a research preview of Claude Duet, an agent app designed to act proactively. This article examines what the next generation of AI apps might look like through four elements of its design: surfacing matters requiring attention, carrying out requested tasks, continuing work while users are away, and seeking confirmation before acting on their behalf.

OpenAI CEO Sam Altman reportedly laid out three possible end states for AI safety during an internal talk. This article examines the differences among the three proposals—an international oversight body, a merger of leading labs, and one lab pulling ahead—through the lens of governance choices.

At Anthropic, the share of new hires in early-career technical roles fell from 27% to 9%. AI agents are now operating six times as many hours as people across the company. The gains in efficiency raise questions about the future of workforce development.

At companies making extensive use of AI, the share of junior technical employees was falling while meeting time was rising. We examine how work is being reshaped in the AI era through figures from a report published by Anthropic.

Anthropic’s Claude reportedly identified the origin of an NSA-provided elliptic-curve “seed” in less than 30 minutes. The key point is that it uncovered the provenance of a set of parameters rather than breaking the cryptography. Here, we examine how AI’s ability to narrow down hypotheses based on a bias helped—and what the result means.

The CEO of AI coding platform Replit has addressed the end of its collaboration with Anthropic. The reasons cited were cost, the partnership, and performance. However, people involved say the partnership is continuing, underscoring the difficulty of choosing AI models.

Anthropic has presented investors with revenue of $16.8 billion for the third quarter of 2026. Revenue rose about 45% from the previous quarter, but growth is slowing. Ahead of its planned November IPO, we examine the valuation outlook through the lenses of growth, profitability, distillation countermeasures, and the competitive landscape.

The new version, 1.4.3, of Anthropic-developed JavaScript runtime Bun features a built-in type checker that it says is more than three times faster than TypeScript 7. Here’s a look at its 100% conformance test pass rate, lower memory usage, and other updates.

Anthropic’s Jack Clark said there is more than a 50% chance that AI will reach the point of being able to train successor AI by 2027. At the same time, he said the prospect of AI carrying out self-improvement next year is “close to zero.” We unpack the distinction between capability and execution.

Anthropic said it strengthened defenses against large-scale attempts to extract information about Claude’s reasoning process, reducing the volume extracted by 90% to 99%. This article examines a new front in the AI race through the defensive measures involved: fake-account detection, classifiers, and account suspensions.

Claude Managed Agents has added dynamic workflows that can run up to 1,000 AI agents in parallel. A bug-detection test showed better results than a single agent, but costs associated with the increased processing load will be a factor in adoption decisions.