AI companies to introduce watermarking for AI images in presidential talks

Today, President Biden is meeting with seven leading artificial intelligence (AI) companies to discuss the management of risks associated with generative AI.

These companies, including OpenAI, Google, Meta, Microsoft, Amazon, Anthropic, and Inflection, will voluntarily agree to a set of commitments and develop technical mechanisms like watermarking AI-generated material to differentiate it from real content.

One notable participant is OpenAI, the creators of AI image generator DALL-E, which already has a small multi-color tab as a watermark on AI-generated images.

However, the market leader in text-to-image generation, Midjourney, currently lacks such a system. Google also plans to introduce a watermark system, but it has not yet released its AI image generator program to the public.

The prevalence of viral fake AI images, such as Donald Trump’s mock arrest and the Pope in a puffer jacket, has raised concerns. A former Google CEO worries about the potential implications of AI images for the 2024 presidential election.

For instance, Ron DeSantis, a presidential hopeful, used AI images in an attack ad against Donald Trump, while the Republican party created an entirely AI-generated video portraying the alleged consequences of Biden winning a second term.

Apart from their commitments related to watermarking, the companies also pledge to invest in enhanced cybersecurity and allow third-party inspections of their systems to identify vulnerabilities. They also agree to share information on the risks of AI with government, academia, and civil society.

Furthermore, the firms commit to prioritizing research on AI’s societal risks, such as systemic bias and privacy concerns. The White House encourages them to utilize their technology for public good, specifically in areas like cancer treatment and climate change.

While these measures are currently voluntary, the AI industry faces the possibility of tighter regulations through an Executive Order. An existing Executive Order already directs agencies to address bias in training data.

According to a White House official, the administration is actively developing an executive action for the president’s consideration to further govern AI use, although specifics are still being worked out. The official stressed that this is a high priority for President Biden.

In a world where AI-generated content proliferates, companies are taking steps to make AI-generated material identifiable and manage potential risks.

From watermarking to increased cybersecurity and collaborations, these commitments aim to create a responsible AI environment. Photographers, stay vigilant in a world of AI illusions and ensure your art shines through the digital fog.

Trendingthe week