OpenAI, Anthropic, and Google Discuss Strengthening AI Security Aspects

OpenAI, alongside Anthropic and Google DeepMind, has been actively involved in discussions regarding the security aspects of artificial intelligence (AI) in recent weeks.

According to a report from TechCrunch on Tuesday (15/9) local time, this was stated by OpenAI’s Head of Global Policy, Chris Lehane. Lehane was reportedly in Washington, USA, to discuss the significant risks posed by AI development with members of the United States Parliament.

These discussions took place after Anthropic CEO Dario Amodei called for AI industry players to cooperate in slowing the development of frontier AI to mitigate security risks. Amodei’s call received support from several AI industry figures, including OpenAI CEO Sam Altman and Google DeepMind CEO Demis Hassabis.

Altman previously stated that OpenAI would work with Anthropic to involve third parties in evaluating their AI systems from a security perspective. Lehane subsequently confirmed that OpenAI, Anthropic, and Google DeepMind have indeed been conducting talks regarding AI security over the past few weeks.

This cooperation between competing companies in AI development has previously raised anti-monopoly concerns, as excessive coordination could potentially affect competition. Amodei had previously proposed limited government exemptions that would allow AI companies to coordinate specifically to handle security risks. However, according to Lehane, the three companies do not require such exemptions to conduct discussions regarding AI security.

Previously, OpenAI, Anthropic, and Google DeepMind were reported to be discussing the formation of a standard-setting body for the AI industry. In July, Hassabis also called for the establishment of a new standard-setting agency in the United States to act as an overseer for AI development. The proposed agency would have the authority to test the most advanced AI models and assist in coordinating industry-wide mitigation steps if security risks increase.

On the other hand, the administration of US President Donald Trump holds a different view regarding the tightening of AI regulations. Trump previously dismissed concerns regarding AI security, arguing that slowing technological development could cause the United States to fall behind China. Trump’s AI advisor, David Sacks, also assessed that concerns regarding the risks posed by AI are being greatly exaggerated.