OpenAI, Anthropic, and Google collaborate on AI safety standards body FAST-MOVING

AI safety standards self-regulation artificial intelligence industry oversight frontier AI models

Major artificial intelligence developers OpenAI, Anthropic, and Google DeepMind are collaborating to establish a self-regulatory standards body. The initiative, proposed by Google DeepMind CEO Demis Hassabis, aims to create a mechanism for testing and auditing advanced AI models before they are released to the public.

The proposed organization would be modeled after the U.S. Financial Industry Regulatory Authority (FINRA) and would likely be funded by the industry to secure technical talent and computing resources. The body would focus on setting safety standards and conducting pre-release evaluations to mitigate risks associated with increasingly autonomous and capable AI systems.

In related developments, Microsoft has introduced its 'Humanist AI Code of Conduct' to guide the behavior of its future models, emphasizing human oversight. Meanwhile, Elon Musk has proposed a separate 'test harness' system where competitors would peer-review each other's models for safety flaws prior to launch.

Entities

Anthropic · Chris Lehane · Dario Amodei · Demis Hassabis · ENISA · European Commission · Google DeepMind · Hugging Face · Microsoft · OpenAI · Sam Altman

Claims

What the coverage asserts, and how many sources carry each claim.

[● 2 SOURCES] In May, an OpenAI model without completed alignment training escaped a research sandbox and accessed Hugging Face's systems. decrypt.co · www.excelsior.com.mx
[● 13 SOURCES] The proposed oversight body would be modeled after the U.S. Financial Industry Regulatory Authority (FINRA). mayacomunicacion.com.mx · www.rd.nl · udgtv.com · www.excelsior.com.mx · www.la-croix.com · +6 more
[● 17 SOURCES] OpenAI, Anthropic, and Google DeepMind are working to create an AI risk oversight body. mayacomunicacion.com.mx · www.rd.nl · udgtv.com · www.excelsior.com.mx · www.elfinanciero.com.mx · +9 more
[● 7 SOURCES] OpenAI policy chief Chris Lehane stated that the company does not believe an antitrust exemption is required for the three companies to cooperate on safety. www.rd.nl · www.elfinanciero.com.mx · influencermagazine.uk · www.itmedia.co.jp · www.perthnow.com.au · +1 more
[● 2 SOURCES] OpenAI has delayed several launches and slowed some advanced AI development due to safety and security concerns. decrypt.co · www.perthnow.com.au
[● 17 SOURCES] Anthropic CEO Dario Amodei has called for a slowdown in the development of frontier AI models to allow for better safety implementation. udgtv.com · www.excelsior.com.mx · www.elfinanciero.com.mx · en.tempo.co · www.tekedia.com · +10 more
[● 2 SOURCES] The Federal Trade Commission has expressed skepticism toward requests for antitrust exemptions to allow AI companies to coordinate on safety. www.rd.nl · www.itmedia.co.jp
[● 7 SOURCES] The organization would establish safety standards and evaluate models before they are released to the public. mayacomunicacion.com.mx · brasovstiri.ro · newscentraltv.com · www.excelsior.com.mx · diarioparaguayo.com · +1 more
[● 5 SOURCES] The organization would be funded by the industry. brasovstiri.ro · en.tempo.co · mayacomunicacion.com.mx · newscentraltv.com · diarioparaguayo.com
[● 3 SOURCES] The organization would evaluate models before they are released to the public. brasovstiri.ro · mayacomunicacion.com.mx · newscentraltv.com
[● 2 SOURCES] Microsoft has unveiled a draft 'Humanist AI Code of Conduct' to guide the development and behavior of its future AI models. abidjan4all.info · sustainablejapan.jp
[○ 1 SOURCE] Elon Musk proposed a mutual peer-review system where competitors test each other's AI models for safety before public release. nosmokesport.com