Skip to content
Neural Daily

BusinessNews

Anthropic says AI may soon speed up its own model R&D

🇭🇺 Olvasd el magyarul →

Anthropic's amended IPO filing says the company may soon reach the threshold where AI substantially accelerates its own development. None of its current models is above it yet.

By: Mikola Máté Patrik · · 2 min read

Anthropic logo
Image: Anthropic · Press kit · Official social image of the Anthropic S-1 announcement · Source

What happened?

On November 2, Anthropic filed an amended version of its IPO prospectus, an S-1/A, with the U.S. Securities and Exchange Commission (SEC). A new paragraph in the risk factors says that, based on internal estimates and expectations about future models, Anthropic may imminently meet the threshold in its Responsible Scaling Policy (RSP) for "dramatically accelerated progress attributable to AI."

This is what the field calls recursive self-improvement (RSI), the framing TechCrunch used too (TechCrunch). According to the filing, Anthropic has not designated any of its current models as exceeding the threshold and is still assessing the data. The company also says it has in the past failed to identify a significant acceleration until after it happened, because its measurements rely on lagging indicators.

Humans stay in the loop

Anthropic says it does not expect to be able to safely use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision, if that ever becomes technically possible. It is therefore considering RSP updates to keep humans in control of frontier model development, and promises further reports on what meeting the threshold implies.

The numbers

The filing puts revenue for the quarter ended September 30 at $17.0 billion, up 1,127% year over year, while compute and infrastructure expenses grew 382% over the same period. Anthropic has about $895.4 billion in commitments, almost all for compute and infrastructure, with $69.8 billion due in the next 12 months. The Decoder, citing Bloomberg and The Information, reports a quarterly net loss of $23.3 billion (The Decoder).

Why it matters

An S-1 is both a mandatory risk disclosure and a document aimed at investors, as TechCrunch notes. Still, a statement in a securities filing carries legal weight, which makes it heavier than a blog post. The next milestone is the RSP update, which will show exactly what human oversight Anthropic commits to in its own model development.

    Produced with the help of AI, reviewed by an editor. How we make our articles

    Reviewed and edited by the Neural Daily desk. About

    Related articles