Anthropic
OpenAI said it was a complex engineering challenge to design deep access for external inspectors without compromising sensitive architecture, and it required meticulous planning.
The friction erupted after Amodei released an exhaustive manifesto calling for a coordinated pause on capabilities, independent model vetting and closer alignment with global governments.
The internal pushback suggests logistical hurdles of transforming public rhetoric into practical engineering workflows, according to a report by The Financial Times. The report says that employees at the development hubs behind ChatGPT and Claude are scrambling to interpret directives from Sam Altman and Dario Amodei, who urged a deceleration in model development following dire warnings from industry researchers. High-profile pledges by the CEOs of OpenAI and Anthropic to slow down the pace of artificial intelligence (AI) development have triggered widespread internal friction, as staff members raise alarms over cybersecurity, proprietary secrets and day-to-day operations. The Anthropic CEO also requested legal exemptions from American antitrust legislation to facilitate cross-lab coordination. The sudden announcement reportedly took rank-and-file employees and specialised safety researchers by surprise. Altman publicly supported the blueprint as a great concept, signaling that OpenAI would adopt similar protocols. Engineering staff worry that handing broad credentials to third parties creates massive vulnerabilities around cutting-edge intellectual property. Teams say these systems are at the center of intense commercial competition and that allowing external access without oversight poses a major security risk.