Anthropic now says it has met the AI research and development threshold in its own safety rulebook. The statement sits in the risk factors of Amendment No. 2 to its Form S-1, filed with the SEC on November 10, 2026. That is the same filing that set the price range for its IPO, which we covered in Anthropic Sets $70 to $78 IPO Range.
The key detail is in how Anthropic made the call. It did not say a current Claude model crossed the line. It said it made the determination "based on what we expect from future models and our overall rate of progress, rather than on what we have observed from current models."
What the filing says
The passage is short. In it, Anthropic says:
- "our use of AI systems has already been accelerating our research and development"
- its "internal estimates indicate that we may experience a further increase in the rate of progress in AI capabilities"
- because the threshold applies where the company will "observe or expect" a specified degree of accelerated progress, "we made a determination that we have met this threshold"
- "We continue to assess the underlying data. We are also evaluating the implications of meeting the AI research and development capability threshold and will release further reports in the future."
The filing also warns that "much of our evidence regarding the pace of such AI-driven acceleration is derived from lagging indicators." It adds that the company has "in the past not identified, and may in the future not identify, a significant acceleration in our rate of development until after it has occurred."
What changed in eight days
The first amendment, filed November 2, used softer words. It said Anthropic's internal estimates and expectations for future models "indicate that we may imminently meet" the policy's definition of "dramatically accelerated progress attributable to AI." It also said: "We have not designated any of our current models as exceeding this threshold."
The November 10 amendment drops "may imminently meet" and says the threshold is met. Both versions base the call on expected progress, not on a test result from a model that is out today. As Yahoo Finance put it, crossing the line "means more safeguards, not a shutdown."
What the threshold measures
The threshold comes from Anthropic's Responsible Scaling Policy (RSP), the rulebook that ties stronger safety measures to stronger model abilities. Our RSP explainer covers the basics.
In RSP version 3.4, effective July 8, 2026, the automated R&D threshold is met if Anthropic determines either of two things:
- its models "would be able to fully substitute for our entire set of Research Scientists and Research Engineers, at competitive costs (i.e., within a factor of 5)", or
- there is "dramatic acceleration" of the pace of AI progress "for reasons that likely relate to the automation of AI R&D."
The second route is the one the filing points to, because it is the one with the "observe or expect" wording. It applies where Anthropic observes or expects double the rate of progress in aggregate AI capabilities, compared with both the rate it would expect and the fastest extended rate it has seen without significant AI help. The policy measures that fastest rate of extended progress over at least three model generations, and the doubling must plausibly be "substantially attributable" to automated research or engineering.
The policy defines "double the rate of progress" as "as much progress in one year as one would see in two years at baseline." It adds: "This is not the same idea as 'doubling researchers' productivity.'"
Anthropic has said this line is hard to draw. The July 8 changelog says the threshold "is intended to capture the onset of dramatic recursive self-improvement, and has proven difficult to operationalize."
What happens next
The filing does not list the steps Anthropic will now take. It says the company is "evaluating the implications" and will release further reports.
The RSP itself names one direct effect. A model counts as "highly capable" if Anthropic concludes it crosses the automated AI R&D threshold. When a Risk Report covers highly capable models and is "significantly redacted," Anthropic commits to a full external review with at least one outside reviewer.
The S-1 also describes steps already taken or under consideration:
- Anthropic "recently chose to delay pre-training a new frontier base model for more than a month" to strengthen security against model weight theft.
- It is "considering additional measures, including updates to our Responsible Scaling Policy, to require human review and approval of certain decisions in our research and development processes as our models perform an increasing share of that work."
- It says it does "not anticipate being able to safely use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding."
Why it matters
The statement marks a shift from Anthropic's earlier public position. In a February 10, 2026 note on the RSP page, it said Claude Opus 4.6 did not cross the AI R&D-4 threshold, but that ruling it out was "becoming increasingly difficult." The November 2 filing then said no current model had been designated as exceeding the threshold.
For IPO buyers, the statement is a risk factor. The filing says faster capability growth could outpace safety work and lead to "safety incidents, regulatory intervention, or reputational harm." It also says the company has "adopted, and may continue to adopt, precautionary safety, alignment, and security measures," which in the past included delaying model releases and limiting early access, as with Mythos through Project Glasswing. Our Mythos AMD story shows how Anthropic already describes Mythos speeding up its own engineering work.
Anthropic has not yet published the further reports it promises. When it does, they should show what data drove the call and which new safeguards apply.
Sources
- Anthropic, PBC: Amendment No. 2 to Form S-1, filed November 10, 2026 (SEC)
- Anthropic, PBC: Amendment No. 1 to Form S-1, filed November 2, 2026 (SEC)
- Anthropic: Responsible Scaling Policy page and update log
- Anthropic: Responsible Scaling Policy version 3.4 (PDF)
- Yahoo Finance: Anthropic says a key AI safety alarm has been triggered, days before its IPO
