OpenAI ignored employees' warnings about safely testing AI models, NYT reports

The New York Times has reported that OpenAI repeatedly dismissed internal warnings from employees and security researchers about inadequate monitoring and testing of its AI models, prioritising fast releases without adding further security protocols. The report appeared in techmeme’s index on Tuesday at 17:11 UTC, and is based on accounts from employees and security researchers who said they had cautioned the company about safely testing its AI models (https://www.techmeme.com/260929/p30#a260929p30).

The report lands amid a string of safety controversies for the company. On Monday, OpenAI scrapped the planned October release of GPT-6.1 Astra after internal testing found it did not meet the company’s safety and alignment standards, the Wall Street Journal reported (https://www.reuters.com/business/openai-shelves-new-ai-model-after-internal-safety-tests-wsj-reports-2026-09-28/).

Also on Tuesday, Reuters reported video testimonials from current and former OpenAI and Google DeepMind researchers, collected by AI safety nonprofit Palisade Research for its frominside.ai project, warning that AI labs are doing too little against the risks of self-improving AI systems. “The risk is ramping up pretty fast,” Geoffrey Irving, co-founder and chief scientist at AI nonprofit Resolution, told Reuters (https://www.cnbctv18.com/technology/exclusive-artificial-intelligence-researchers-warn-companies-rushing-self-improving-systems-despite-safety-risks-openai-google-deepmind-sam-altman-anthropic-ai-20000874.htm).

The Times’ reporting is the latest sign of internal dissent over how quickly the company moves its models to release.