OpenAI and Anthropic Seek Greater Control Over External AI Safety Reports

Artificial intelligence developers OpenAI and Anthropic are pushing for increased control over what external auditors can publish in independent safety reports. Windows Report, PYMNTS.com, and The420.in all reported that the artificial intelligence companies are seeking greater oversight over public findings. Additionally, Tech Times and Crypto Briefing reported that both organizations pushed for veto rights regarding evaluator publications, citing concerns over trade secrets. [1][5][6][7][8]

The scrutiny over safety reviews comes alongside legislative discussions regarding industry governance. Wired AI reported that executives from four frontier artificial intelligence firms, alongside whistleblowers, testified before the Business, Innovation, Science, and Trade Committee of the British Parliament. During the session, Anthropic head of Europe Liam Booth-Smith stated that rules governing safety should be binding. [3]

Industry executives are also meeting internationally to address AI safety concerns. The Irish Star reported that representatives from OpenAI, Anthropic, and Microsoft gathered in Dublin for an AI safety summit. [2]

OpenAI and Anthropic are both public benefit corporations headquartered in San Francisco that develop frontier generative artificial intelligence models. AI safety research focuses on preventing accidents, misuse, and alignment risks associated with advanced artificial intelligence systems. [9][10][11]