Continue Anthropic’s model training rather than stop unilaterally while competitors advance; this does not reject coordinated future pacing tools. [1]
Changed circumstances. The original interview expressly defends continued model training under competitive conditions. No call for increased speed appears in the verified passage. July collective pacing-tools support is narrower than a recommendation to slow now; exact individual signature date remains unknown. No corporate policy is automatically attributed to Kaplan.
Source [1] · Original interview
Anthropic Drops Flagship Safety Pledge
Attributed to: Jared Kaplan · time.com
Published
“We felt that it wouldn't actually help anyone for us to stop training AI models.”
Body paragraph beginning “We felt that”; direct interview quotation attributed to chief science officer Jared Kaplan · View original source ↗
Source context. Original reporting of an exclusive Kaplan interview about changing Anthropic’s Responsible Scaling Policy. Publication date is known; the exact interview event date is not disclosed.
Scope
Anthropic’s model training while competitors continue frontier development
Conditions
Continuation is defended against a unilateral halt while competitors advance; coordinated future interventions are not excluded.
Stated rationale
The original interview expressly defends continued model training under competitive conditions. No call for increased speed appears in the verified passage. July collective pacing-tools support is narrower than a recommendation to slow now; exact individual signature date remains unknown. No corporate policy is automatically attributed to Kaplan.
Source details
primary · Independent original-source verification; no Grok attribution.
Discovered 3 Oct 2026.
Published 24 Feb 2026.
Accessed 3 Oct 2026.
Editorial assessment. The original interview expressly defends continued model training under competitive conditions. No call for increased speed appears in the verified passage. July collective pacing-tools support is narrower than a recommendation to slow now; exact individual signature date remains unknown. No corporate policy is automatically attributed to Kaplan. Research draft only. Independent adversarial review and exact-record human approval remain required.
Assessment reviewed
Earlier position ·
PACE
Anthropic’s 2023 policy ties frontier scaling to required safety procedures, including temporary pauses if scaling outstrips safeguards. [1]
Scope distinction. Historical 2023 company-specific safety-conditioned scaling policy, distinct from the 2026 defense against unilateral cessation.
Source [1] · Original statement
Jared Kaplan written statement to Senate AI Insight Forum
Attributed to: Jared Kaplan · www.schumer.senate.gov
Statement
“…temporarily pause training of more powerful models if our AI scaling outstrips our ability to comply with the necessary safety procedures.”
PDF page 5 of 6, Anthropic’s Responsible Scaling Policy section, paragraph beginning On the one hand · View original source ↗
Source context. Authored written Senate statement by Jared Kaplan, Anthropic PBC, December 6 2023. Corporate policy explanation is explicitly organizational attribution and preserved as historical scope, not silently generalized to an industry-wide personal pause.
Scope
Anthropic frontier-model training under its 2023 Responsible Scaling Policy
Conditions
Temporary pause if scaling outstrips required safety procedures; solve safety issues to unlock further scaling.
Stated rationale
Authored written Senate statement by Jared Kaplan, Anthropic PBC, December 6 2023. Corporate policy explanation is explicitly organizational attribution and preserved as historical scope, not silently generalized to an industry-wide personal pause.
Source details
primary · Independent original-source verification; no Grok attribution.
Discovered 3 Oct 2026.
Accessed 3 Oct 2026.
Editorial assessment. Historical organizational scope preserved. Not an industry-wide personal stance or explicit revision.