If You Can't BEAT Apple…

Figma pulls its generative AI tool… YouTube will take down deepfakes… and Meta kills the “Made with AI” tag

Figma Disables AI Design Tool After Copying Apple

If you or your colleagues opened up the design tool Figma this morning to get to work, you may have noticed one big piece missing: Its new much-ballyhooed generative AI tool has been cut from the app.

It’s a temporary cut, and for a little bit of an embarrassing reason: It was found to be copying Apple’s weather app almost to the pixel.

If you can’t beat ‘em…

One designer asked Figma for mockups for a generic mobile weather app, then asked for another, then another — and each time, Figma would just reproduce essentially the Apple app: widgets in the same position, data in the same order, graphical presentation almost identical.

Figma’s CEO acknowledged the problem, saying he’d been pushing the team too hard to meet a conference deadline. He said the AI feature would be temporarily disabled until it could produce reliable outputs.

How did it get in there?

As for how Apple’s weather app got lifted in training, that’s a little more murky.

Figma uses multiple vendors for its AI models, including OpenAI, Amazon Titan, and Jasper. These vendors do not disclose their training data.

Legal consequences unlikely

Sites Hit by September SERP Update Sneak Up

Sites hit by Google's Helpful Content Update which rolled out last September are seeing small improvements in their search ranking position over the past week.

Small — nothing like full recoveries, though some people remain hopeful.

Barry Schwartz at SERoundtable says:

This noted uplift came from a small study of 384 sites which had been heavily impacted. They’re still lower than before the September update, but they have been sneaking up, on average.

Then again, there was a small algorithm update to Google’s search engine over the weekend, so it’s possible that played a role as well.

YouTube Will Take Down Deepfakes — If You Ask

In May, Scarlett Johansson threatened OpenAI with legal action for using an AI-generated version of her voice for ChatGPT 4.o. The CEO denied the accusation.

Now, Johansson might be pleased to see YouTube letting users request the removal of AI-generated content that resembles them.

Humans will likely review

And it seems YouTube will do this with human eyes. It says the person involved must be “uniquely identifiable,” and they’ll also consider if the content is satire, parody, or involves a public figure in a sensitive situation.

YouTube says it will only accept first-party claims, meaning agencies or legal firms won’t be able to file the request on behalf of their clients.

What happens after filing

After filing a request, the content won't be immediately removed.

The uploader will have 48 hours to trim, blur, or remove the video. Making a video private won’t be enough. If the uploader doesn't act within 48 hours, YouTube says it will take further action if needed.

Meta Tweaks the “Made with AI” Tag

Meta yesterday announced it’s updating that Made with AI label it applies to images uploaded to its apps — it will now read AI info instead.

The change comes after complaints that the label was being slapped onto images even if they weren’t made with AI — even if all that happened was a little machine learning had been involved in a retouch or a compression.

The new label is rolling out first on mobile apps and later on the web.

Clicking the tag will still provide a detailed explanation of why it was applied, covering images both fully generated by AI or edited with AI tools.

– 30 –