A number of AI companies, notably OpenAI, Alphabet, and Meta Platforms, have made voluntary commitments to the security of their products. At a White House event intended to allay worries about potential AI abuse and its effects on US democracy, US President Joe Biden announced those pledges.
While praising these commitments as a significant milestone, Biden underlined that there is still much teamwork to be done. To protect national security and democratic ideals, he emphasized the need to be attentive about the dangers posed by developing technology, including AI.
Anthropic, Inflection, Amazon, and Microsoft are also among the businesses taking part in these voluntary pledges. They have committed to thoroughly testing AI systems before they are made public, sharing knowledge about risk mitigation strategies, and making investments in cybersecurity to guard against future consequences.
The seven companies have agreed to create a watermarking system that can be used on all types of AI-generated content, including text, photos, audio, and videos, as part of their pledges. Users will be able to tell when AI technology was utilized to create material because the watermark will technically be integrated in it.
The goal of this watermarking endeavor is to make it easier for users to spot deep-fake audio or visuals that could be used to perpetrate fraud, portray fictitious violence, or defame politicians. Uncertainty persists regarding the mechanics of how the watermark will be visible when information is shared.