OpenAI, Google DeepMind and Anthropic are discussing the possibility of creating a new industry body focused on developing common safety standards for advanced artificial intelligence.
The discussions come as concerns grow over the rapid development of increasingly powerful AI systems and the potential risks associated with deploying them without sufficient safety testing. The companies remain fierce competitors, but they are exploring whether cooperation on AI safety could help establish clearer standards across the industry.
Proposed AI Safety Body
The idea gained momentum after Google DeepMind chief Demis Hassabis proposed an industry-wide standards organisation earlier this year.
His proposal was compared with the Financial Industry Regulatory Authority, or FINRA, which oversees parts of the US financial sector. A similar organisation for AI could potentially evaluate advanced models before they are widely released.
The proposed body could involve government oversight, industry funding and independent technical experts. It could also include representatives from the open-source AI community.
However, the exact structure and responsibilities of such an organisation have not yet been decided.
Why AI Companies Want Common Standards
Artificial intelligence is advancing quickly, with new models gaining stronger reasoning, coding, research and autonomous capabilities.
As these systems become more powerful, companies and policymakers are increasingly concerned about cybersecurity threats, misuse and other potentially serious consequences.
A common testing framework could allow companies to evaluate advanced AI models using similar safety benchmarks before releasing them to users.
OpenAI chief scientist Jakub Pachocki has also said that shared safety standards and international coordination should be priorities as AI development continues.
The companies have reportedly been discussing possible standards with outside organisations, with further details expected in the coming months.
No Final Agreement Yet
Despite the discussions, there is currently no final agreement establishing the proposed standards body.
There is also disagreement within the technology industry about how AI should be regulated and who should be responsible for overseeing powerful AI systems.
Some technology leaders support stronger safeguards, while others have raised concerns that excessive regulation could slow innovation or give established companies an advantage over smaller competitors.
The US government is also considering its own approach to AI oversight, adding another layer to the ongoing debate over how advanced AI should be tested and regulated.
What Could Change for AI Development?
If the initiative eventually becomes a formal organisation, it could create a more consistent system for assessing advanced AI models before they reach the public.
Such standards could cover areas including model safety testing, security evaluations, risk assessments and independent audits.
For users and businesses, common standards could make it easier to understand how different AI systems have been evaluated.
However, creating rules that are accepted by major AI companies, governments and independent researchers could be difficult.
AI Safety Becomes a Bigger Industry Issue
The discussions between OpenAI, Anthropic and Google come at a time when several AI leaders are calling for greater attention to safety.
Anthropic CEO Dario Amodei has recently argued that AI development needs to slow enough for safety measures to catch up. OpenAI CEO Sam Altman and other technology leaders have also expressed support for stronger safety efforts.
For now, the proposed standards body remains an idea under discussion rather than an established regulator.
If the leading AI companies can agree on meaningful and independent safety standards, however, the initiative could become an important step toward making the development of advanced artificial intelligence more transparent and responsible.