In a significant move to address AI safety concerns, Big tech companies pledge to adopt voluntary guidelines, such as watermarking to help users identify AI-generated content, as part of their efforts to check AI abuse and misuse. These companies, Amazon, Anthropic, Google, Meta, Microsoft, OpenAI and ML startup Inflection agreed with the Biden Administration to submit their systems to testing by independent experts. This move comes amid the popularity and investment in generative AI technology, prompting global lawmakers to consider regulations to safeguard national security and the economy.
Addressing concerns over AI technology
The rapid rise of generative AI, which uses data to create human-like content, has raised concerns about its potential risks to various domains. U.S. Senate Majority Leader Chuck Schumer called for comprehensive legislation to ensure appropriate safeguards on artificial intelligence in June. To address these concerns, Congress is contemplating a bill requiring political advertisements to disclose whether AI was used to generate content.
President Joe Biden is actively working on developing an executive order and bipartisan legislation to regulate AI technology further. As part of this effort, executives from the seven major AI companies gather at the White House on Friday to discuss their commitments. The companies agreed to develop a watermarking system for all forms of AI-generated content, including text, images, audio, and videos. This watermarking technique aims to provide users with a clear indication of when AI technology has been used to create content.
Watermarking for enhanced AI safety
The proposed watermarking system is expected to be essential in combating the potential dangers associated with AI-generated content. By technically embedding a watermark into the content, users will be better equipped to identify deep-fake images or audios that may depict non-existent violence, facilitate scams, or present politicians in a bad light. However, details about how the watermark will be evident during information sharing have not been clarified.
Digital watermarks verify the authenticity or integrity of the carrier signal or show the identity of its owners. Not only does a watermark dissuade individuals from leaking documents, but if a leak occurs, the source of the leak can be easily identified when a watermark with the allowed recipient’s name is placed on the file. Digital watermarking is a technology in which identification information is embedded into the data carrier in ways that cannot be easily noticed, and in which the data usage will not be affected. This technology often protects copyright of multimedia data, and protects databases and text files.
Commitments beyond watermarking
The companies’ commitments go beyond watermarking and extend to protecting users’ privacy as AI technology develops. They have also vowed to ensure that their AI systems are free of bias and refrain from being used to discriminate against vulnerable groups. The companies said they will publicly report the capabilities and limitations of their AI systems, as well as guidelines for its ethical use. The companies plan to conduct research on societal risks posed by AI. Additionally, these companies have pledged to deploy AI solutions to address scientific challenges, such as medical research and climate change mitigation.
A step forward for AI safety
The voluntary commitments made by these major AI companies signify a significant step forward in addressing the safety concerns associated with AI technology. With the cooperation of industry leaders and ongoing regulatory efforts, there is optimism that AI can continue to flourish while adhering to responsible practices.
As generative AI continues gaining popularity and investment, regulators and lawmakers worldwide have focused on addressing potential risks. The commitments made by top AI companies to implement watermarking measures and enhance safety through responsible practices mark a positive development in the field. President Biden’s efforts to develop executive orders and bipartisan legislation further reinforce the importance of ensuring AI technology’s safe and ethical use. AI can unlock its potential while mitigating societal risks with continued collaboration between industry stakeholders and regulatory bodies.
