Seven U.S Tech Companies Voluntarily Commit To AI Guardrails
Companies that volunteered for AI safeguards include Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI.
- Late last week, the White House confirmed that Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI signed up to self-regulate their AI developments.
- The seven companies signed a document titled Ensuring Safe, Secure, and Trustworthy AI, which stands on three pillars: safety, security, and trust.
Seven companies engaged in artificial intelligence (AI) development have volunteered to self-regulate their tech. Late last week, the White House confirmed that Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI are signatories to this document to ensure the “safe, secure, and transparent development of AI technology.”
Generative AI has taken the world by storm in the last nine months. The Summer 2023 Fortune/Deloitte CEO Survey revealed that 37% of respondents, CEOs from 19 industries, said their companies are already implementing generative AI to some degree. 55% of CEO respondents said they are evaluating & experimenting with generative AI.
An even higher number, 79%, believe generative AI will help improve efficiency, while 52% said it will increase growth opportunities. However, it isn’t a stretch to term AI as a double-edged sword, considering its security and privacy ramifications, not to mention its tendency to spew mis or disinformation.
This is why the document titled Ensuring Safe, Secure, and Trustworthy AI, signed by the seven companies, stands on three pillars: safety, security, and trust.
How Will Companies Self-Regulate AI Development?
Safety
As part of the safety protocol, companies have committed to red-teaming efforts to eliminate societal risks and national security concerns, such as the tech’s applicability to developing biological, chemical, and radiological weapons; cybersecurity risks (vulnerability discovery), bias or discrimination, and the risk of self-replication.
The companies have also committed to information-sharing efforts among each other and with the government. The signatories would have to participate in a forum or establish a mechanism that oversees developing, advancing, and adopting shared standards and best practices for frontier AI safety.
See More: OpenAI Faces Its First Serious Regulatory Turbulence Over ChatGPT
Security
Under security commitments, companies agreed to establish external and internal threat detection programs. They would also incentivize third-party vulnerability detection and responsible disclosure through bug bounty programs, contests, or prizes.
Trust
To ensure trust, the company would institute provenance and/or watermarking systems for any audio or visual content created by any of their proprietary and publicly available AI tools. The document necessitates periodic safety evaluations complete with the capabilities, limitations, and details of what constitutes appropriate and inappropriate use for all AI service versions.
Finally, signatories would also need to support research & development initiatives to overcome major societal challenges, including climate change, early cancer detection and prevention, and combating cyber threats.
The document applies only to generative AI tools and models more powerful than the current industry standards. These include GPT-4, Claude 2, PaLM 2, Titan, and, in the case of image generation, DALL-E 2.
Representatives from the seven companies confirmed their commitment to the initiative on their respective blogs. Meta president of global affairs, Nick Clegg, said, “Meta welcomes this White House-led process, and we are pleased to make these voluntary commitments alongside others in the sector. They are an important first step in ensuring responsible guardrails are established for AI, and they create a model for other governments to follow.”
Status of AI Regulations in Other Countries
While a federal law concerning AI regulations takes shape, the European Union is on track to pass its AI Act later this year. Earlier in July, China also released updated guidelines to regulate AI in the country, which is set to take effect on August 15, 2023.
“As we advance this agenda at home, the Administration will work with allies and partners to establish a strong international framework to govern the development and use of AI,” the White House said. It has already consulted on the voluntary commitments with Australia, Brazil, Canada, Chile, France, Germany, India, Israel, Italy, Japan, Kenya, Mexico, the Netherlands, New Zealand, Nigeria, the Philippines, Singapore, South Korea, the UAE, and the UK.”
Nevertheless, Sen. Mark Warner is concerned about China getting a “head start.” He told Politico earlier this month that China is “very much ahead of the game in terms of self-regulating AI within their own nation-state.”
He added that China has “moved even further than Europe in having specific legislation.”
Should companies be allowed to self-regulate their AI developments? Share your thoughts with us on LinkedIn, Twitter, or Facebook. We’d love to hear from you!
Image source: Shutterstock
MORE ON ARTIFICIAL INTELLIGENCE
