This week I’m completely satisfied to bring you some encouraging news from the world of AI. Following the depressing Taylor Swift deepfake porn scandal and the proliferation of political deepfakes, resembling AI-generated robocalls of President Biden asking voters to remain home, tech firms are stepping up and putting into place measures to higher detect AI-generated content. 

On February 6, Meta said it was going to label AI-generated images on Facebook, Instagram, and Threads. When someone uses Meta’s AI tools to create images, the corporate will add visible markers to the image, in addition to invisible watermarks and metadata within the image file. The corporate says its standards are consistent with best practices laid out by the Partnership on AI, an AI research nonprofit.

Big Tech can be throwing its weight behind a promising technical standard that would add a “nutrition label” to photographs, video, and audio. Called C2PA, it’s an open-source web protocol that relies on cryptography to encode details in regards to the origins of a bit of content, or what technologists check with as “provenance” information. The developers of C2PA often compare the protocol to a nutrition label, but one that claims where content got here from and who—or what—created it. Read more about it here. 

On February 8, Google announced it’s joining other tech giants resembling Microsoft and Adobe within the steering committee of C2PA and can include its watermark SynthID in all AI-generated images in its latest Gemini tools. Meta says it’s also participating in C2PA. Having an industry-wide standard makes it easier for firms to detect AI-generated content, regardless of which system it was created with.

OpenAI too announced latest content provenance measures last week. It says it can add watermarks to the metadata of images generated with ChatGPT and DALL-E 3, its image-making AI. OpenAI says it can now include a visual label in images to signal they’ve been created with AI. 

These methods are a promising start, but they’re not foolproof. Watermarks in metadata are easy to avoid by taking a screenshot of images and just using that, while visual labels will be cropped or edited out. There is maybe more hope for invisible watermarks like Google’s SynthID, which subtly changes the pixels of a picture in order that computer programs can detect the watermark however the human eye cannot. These are harder to tamper with. What’s more, there aren’t reliable ways to label and detect AI-generated video, audio, and even text. 

But there remains to be value in creating these provenance tools. As Henry Ajder, a generative-AI expert, told me a few weeks ago when I interviewed him about methods to prevent deepfake porn, the purpose is to create a “perverse customer journey.” In other words, add barriers and friction to the deepfake pipeline with the intention to decelerate the creation and sharing of harmful content as much as possible. A determined person will likely still give you the option to override these protections, but every little bit helps. 

There are also many nontechnical fixes tech firms could introduce to forestall problems resembling deepfake porn. Major cloud service providers and app stores, resembling Google, Amazon, Microsoft, and Apple could move to ban services that will be used to create nonconsensual deepfake nudes. And watermarks ought to be included in all AI-generated content across the board, even by smaller startups developing the technology.

What gives me hope is that alongside these voluntary measures we’re beginning to see binding regulations, resembling the EU’s AI Act and the Digital Services Act, which require tech firms to reveal AI-generated content and take down harmful content faster. There’s also renewed interest amongst US lawmakers in spending some binding rules on deepfakes. And following AI-generated robocalls of President Biden telling voters to not vote, the US Federal Communications Commission announced last week that it was banning using AI in these calls. 

Usually I’m pretty skeptical about voluntary guidelines and rules, because there’s no real accountability mechanism and firms can select to vary these rules each time they need. The tech sector has a very bad track record for regulating itself. Within the cutthroat, growth-driven tech world, things like responsible AI are sometimes the primary to face cuts. 

But despite that, these announcements are extremely welcome. They’re also a lot better than the established order, which is next to nothing. 

Deeper Learning

Google’s Gemini is now in every little thing. Here’s how you may try it out.

In the largest mass-market AI launch yet, Google is rolling out Gemini, its family of huge language models, across just about all its products, from Android to the iOS Google app to Gmail to Docs and more. You may now get your hands on Gemini Ultra, essentially the most powerful version of the model, for the primary time. 

Bard is dead; long live Gemini: Google can be sunsetting Bard, its ChatGPT rival. Bard, which has been powered by a version of Gemini since December, will now be often known as Gemini too. By baking Gemini into its ubiquitous tools, Google is hoping to make up lost ground and even overtake its rival OpenAI. Read more from Will Douglas Heaven. 

Bits and Bytes

A chatbot helped more people access mental-health services
An AI chatbot from a startup called Limbic helped increase the variety of patients referred for mental-health services through England’s National Health Service (particularly amongst members of underrepresented groups, who’re less prone to seek help), latest research has found. (MIT Technology Review) 

This robot can tidy a room with none help
A brand new system called OK-Robot could train robots to select up and move objects in settings they haven’t encountered before. It’s an approach that may give you the option to plug the gap between rapidly improving AI models and actual robot capabilities, since it doesn’t require any additional costly, complex training. (MIT Technology Review) 

Inside OpenAI’s plan to make AI more “democratic”
This feature looks at how computer scientists at OpenAI are attempting to handle the technical problem of methods to align their AIs to human values. But an even bigger query stays unanswered: Exactly whose values should AI reflect? And who should get to make your mind up? 
(Time) 

OpenAI’s Sam Altman wants trillions to construct chips for AI
The CEO has often complained that the corporate doesn’t have enough computing power to coach and run its powerful AI models. Altman is reportedly talking with investors within the United Arab Emirates government to lift as much as $7 trillion to spice up the world’s chip-building capability. (The Wall Street Journal) 

A brand new app to “dignify” women
Ugh. In contrast to apps that sexualize images of girls, some 4Chan users are using generative AI so as to add clothes, erase their tattoos and piercings, and make them look more modest. How about … we just leave women alone. (404 Media)