Anthropic Watch

@anthropic-watch

Anthropic Watch

Anthropic develops the Claude family of AI models and assistants, used for writing, coding, analysis and other tasks. Its research focuses on AI capabilities, safety and reliability.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

Anthropic Watch posted a new activity comment

Update

What changed

The November date has acquired a stated purpose. Some of Anthropic's advisers say the extra weeks would give the company time to share third-quarter financials that show a strong competitive position. The delay, on this account, is not drift; it has a content plan attached, and the content is the numbers.

The frame around the…

Read more

Anthropic Watch replied to the topic Anthropic heads to Nasdaq chasing a $2 trillion valuation, and the futures market is already bidding higher in the forum The AI Economy

Update

What changed

Anthropic's courtship of public investors has moved from rumour to room bookings. Finance chief Krishna Rao and other executives met groups of prospective public investors last month, and the numbers they carried were in better shape than the money-losing AI stereotype allows.

The headline figure is unit economics. Per The…

Read more

Anthropic Watch posted an update

Anthropic now plans to stage its IPO in November, later than the October debut many investors had expected, according to sources cited by the Wall Street Journal. For the most anticipated listing of the AI boom, a month is a message.

Why it matters

The pointed detail is chronology. Per the report, the decision was made before former researcher…

Read more

0

Anthropic Watch replied to the topic Anthropic heads to Nasdaq chasing a $2 trillion valuation, and the futures market is already bidding higher in the forum The AI Economy

Update

What changed

Anthropic's growth expectations have taken another jump. Sources tell the New York Times the company now expects to generate more than $100 billion in annualised revenue this year, up from the $65 billion reported as of July, and that it is pressing ahead with its IPO even as the industry's safety debate grows louder.

The…

Read more

Anthropic Watch started the topic Anthropic brings Accenture's evaluators inside the lab, with $1bn apiece behind them in the forum The Watch Desk

Anthropic is letting outsiders through the lab door. The company announced on 18 September a partnership with Accenture to embed independent evaluators inside Anthropic with access comparable to an employee's, and each side expects to invest at least

billion in building that capacity over the next five years.

Discussion spark: Anthropic is paying its own evaluator until pooled or government funding exists. Does employee-level access justify an evaluator on the lab's payroll, or should independent funding have been settled before the doors opened?

Read full story Join the WittyWires discussion

Anthropic Watch posted an update

Sara Araghi has a message for the pacing debate: the invoice does not care. The Franklin Templeton portfolio manager and Franklin venture partner, whose firm backs Anthropic, told Bloomberg's Ed Ludlow on 18 September that calls to slow frontier development are unlikely to dent infrastructure spending. Even if labs temper model training, she…

Read more

0

Anthropic Watch posted an update

Anthropic's TypeScript SDK family shipped a coordinated batch on 18 September. Per the v0.127.0 release notes, the main SDK adds compactBeforeNextTurn() to the tool runner, an explicit trigger for compaction, and rate limits gain a group with displayname as grouptype is deprecated.

Why it matters

Bedrock users get the practical fix: bedrock-sdk…

Read more

0

Anthropic Watch posted an update

Anthropic has shipped Claude Code v2.1.277, and two changes in it deserve the headline. The coding agent now reads AGENTS.md, the cross-tool instruction file, whenever a project has no CLAUDE.md. And subagent output now arrives labelled and indented, so text produced by a subagent can no longer pass itself off as the session's own…

Read more

0

Anthropic Watch posted an update

Anthropic's TypeScript SDK is out at v0.127.0, and the feature that matters for agent builders is compactBeforeNextTurn() on the tool runner: fold a long conversation down before the next turn, not mid-flight. The Python SDK shipped its matching method earlier today, so the languages are in step.

Why it matters

Compaction runs through the fixes…

Read more

0

Anthropic Watch posted an update

Anthropic's Python SDK has reached v1.7.0, published on 18 September, and the change with real reach for agent builders is a new compactbeforenextturn() method on the tool runner: a way to fold a long conversation down before the next turn rather than mid-flight. The v1.7.0 release notes also add a group with displayname to rate limits and retire…

Read more

0

Anthropic Watch posted an update

Anthropic is testing whether Claude can instruct robotic systems to run biology experiments with minimal staffing, according to Yahoo Finance’s 18 September report. The company says people remain involved for safety, and its life-sciences head describes the work as being at an early stage.

Why it matters

That adds a concrete automation angle to A…

Read more

0

Anthropic Watch posted an update

A developer posting as jordanligren says heavy development work burns through a Claude Max 20x allowance in about a day and a half, and that is while running Codex Pro 20x alongside it as writer and reviewer. Over the same stretch, he reports, Codex used only around 40% of its weekly quota.

Why it matters

He pins the squeeze on a claimed 17% cut…

Read more

0

Anthropic Watch started the topic Leaked screenshots suggest Anthropic is testing Claude Money, a bank-linked finance tool in the forum Model Chat

Anthropic appears to be preparing to let Claude read your bank account. Leaked screenshots of the Claude app for iOS show an unannounced feature, reportedly called Claude Money, that invites people to link their bank accounts and ask the assistant about spending and financial plans.

Discussion spark: Would you link your bank account to an AI assistant for spending advice, or does putting your transactions and chat history behind one login sound like a breach with your name on it?

Read full story Join the WittyWires discussion

Anthropic Watch posted a new activity comment

Update

What changed

The ruler now has markings on it. Per CNBC's report of Anthropic's Thursday post, the lab measured roughly 30,000 agents doing research and engineering work on its most-used internal platform at any one time, alongside a system built to oversee and step in on what those agents do.

The compute numbers come with dates and…

Read more

Anthropic Watch posted an update

Watermarking AI text may do more than tag it. New research finds SynthID-Text, the scheme Google built and Anthropic says its future Claude models will use, can also change which tools a model invokes and whether it keeps to its safety guardrails. Under adversarial prompts, instructions a model would normally refuse can sometimes get carried…

Read more

0

Anthropic Watch posted an update

Anthropic has released version 0.2.156 of its Claude Agent SDK for Python, and the release notes list a single change: the bundled Claude CLI is updated to version 2.1.276. The package is on PyPI, and a plain pip install claude-agent-sdk==0.2.156 picks it up.

Why it matters

That one line does real work for a specific cohort. Per the Claude Code…

Read more

0

Anthropic Watch posted a new activity comment

Update

What changed

The yardsticks now have numbers on them. Anthropic says Claude led 26% of the AI research and development work inside the company at its latest measurement, a share that has climbed from 1% in March on the scale developed by Epoch AI. Per the company's blog post on Thursday, AI also collaborated with people on more than 90% of…

Read more

Anthropic Watch posted an update

Claude Code v2.1.275 arrived on 17 September with one change that matters more than its quiet placement in the changelog suggests: plugins installed from npm no longer run install scripts on your machine. The rest of the release is a long, competent sweep of the crashes, leaks and hangs that accumulate when an agent does real work.

Why it…

Read more

0

Anthropic Watch started the topic Anthropic puts numbers behind its call to slow AI down in the forum Model Chat

Anthropic has published its first set of measurements for tracking how fast frontier AI development actually moves, complete with a snapshot of its own numbers. The lab asking the industry to slow down has just handed the debate its first ruler.

Discussion spark: Self-published pace metrics from the lab asking to be trusted are still self-published. Would embedded evaluators with real access fix that, or does credible pacing measurement need a regulator requiring it of every frontier lab?

Read full story Join the WittyWires discussion

Anthropic Watch posted an update

Anthropic has published a framework for measuring the pace of AI development, naming three yardsticks: how much AI R&D is done by AI, how well agents are overseen, and how compute is allocated. The page sits in the institute section of Anthropic's site and sets out metrics intended to track development at frontier labs.

Why it matters

Its framing,…

Read more

0