US announces commitment to make AI safer – 7/21/2023 – Tech

The US White House announced this Friday (21) that Amazon, Anthropic, Google, Inflection, Meta, Microsoft and OpenAI agreed to respect several principles in the development of AI (artificial intelligence) to make it safer and more transparent.

These American companies that pioneered this technology have committed themselves to working especially on AI-powered content tagging systems in order to reduce the risks of fraud and misinformation.

They promised to test their programs internally and externally before launching them, invest in cybersecurity and share relevant information about their tools with authorities and researchers, including possible flaws.

They must also “develop sound techniques to ensure users know when content has been generated by AI, such as a watermarking system,” a US government statement said.

“This will allow AI-powered creativity to thrive, while reducing the dangers related to fraud or mistakes,” he added.

Fake photos or extremely realistic manipulated videos, also known as “deepfakes”, have been around for years, but generative AI, capable of producing images at a simple request, raises fears of waves of uninformative content online.

These materials can be used to create very convincing scams or even to manipulate public opinion, a worrying prospect, especially with the approach of the 2024 elections in the United States.

Until now, the major companies involved have been largely reluctant to add such tags to content created with their tools.

“It’s a complicated subject,” a senior White House official admitted to reporters. The watermark “should work for both visual and sound content,” he explained. “Technically, it has to be robust, but also easy for users to see.”

“It’s a good first step in helping the public identify content created with AI,” commented James Steyer, founder of the NGO Common Sense Media.

“However, this type of marking alone will not be enough to prevent malicious actors from using this content for nefarious or illegal purposes”, pondered the expert, mentioning the existence of pirated generative AI programs available on the internet.

In May, the White House stressed the “moral duty” of artificial intelligence companies to ensure the safety of their products.

Political tensions in Congress make it unlikely to legislate on AI at the moment, although the administration has said it is working on an executive order.