AI Firms Will Pledge to Watermark AI Images in Meeting With the President

White House

Seven leading artificial intelligence (AI) firms will meet with President Biden today when they will agree to a set of commitments to manage the harmful risks associated with generative AI.

The Biden administration says that they are voluntary commitments and essentially amount to self-policing by the tech firms.

The companies say they will develop technical mechanisms like watermarking AI-generated material so that people know what they’re looking at is not real.

OpenAI (the makers of AI image generator DALL-E) along with Google, Meta, Microsoft, Amazon, Anthropic, and Inflection are all sending representatives to the White House today where they will discuss the non-binding agreement.

DALL-E already has a watermark on AI-generated images but it is a small multi-color tab on the bottom right-hand side of the image. Text-to-image generator Midjourney, which is the market leader, has no such system.

Google has said it too is developing a watermark system but so far it hasn’t released its AI image generator program to the public.

In recent months, fake AI images have repeatedly gone viral including Donald Trump being arrested and the Pope in a puffer jacket.

A former Google CEO has expressed concern that the 2024 presidential election will be a “mess” because of AI images.

This was after presidential hopeful Ron DeSantis launched an attack ad on main rival Donald Trump where he used AI images to display an alleged cozy relationship with Anthony Fauci and the Republican party made an entirely AI-generated video depicting what will allegedly happen in Biden wins a second term in office.

What are the AI Companies Pledging to?

The firms are also committing to investments in more cybersecurity and will agree to third-party inspections to look for vulnerabilities in their systems

The companies will also commit to sharing information across government, academia, and “civil society” on the risks of AI.

Some of the pledges sound ambiguous, such as reporting AI systems’ “capabilities, limitations, and areas of appropriate and inappropriate use.”

The White House also wants the firms to prioritize research on AI’s societal risks like systemic bias or privacy issues. As well as urging the companies to use their technology for good like fighting cancer and climate change.

As noted by Tech Crunch, while the above measures are voluntarily for now the threat of an Executive Order looms over the AI industry. One is already in force asking agencies to watch out for bias in the training data.

A White House official told reporters on Wednesday that an executive order more strongly regulating AI will eventually be signed but officials are still working out the details.

“The White House is actively developing executive action to govern the use of AI for the president’s consideration,” the official says. “This is a high priority for the president.”

Image credits: Header photo licensed via Depositphotos.