Artificial intelligence is more a part of our lives than ever before. While some might call it hype and compare it to NFTs or 3D TVs, AI is causing a sea change in nearly every part of the technology industry. OpenAI’s ChatGPT is arguably the best-known AI chatbot around, but with Google pushing Gemini, Microsoft building Copilot, and Apple working to make Siri good, AI is probably going to be in the spotlight for a very long time. At The Verge, we’re exploring what might be possible with AI — and a lot of the bad stuff AI does, too.
HP is acquiring the Humane platform and team for $116 million and shutting down all services for its AI Pin devices.
If you’re a Gemini Advanced user, you can now dive deeper into different topics from the Gemini app on Android or iOS. Deep Research was only available on the web when it launched last year.
Topaz Labs is letting users try out its Project Starlight diffusion AI model for enhancing and upscaling video, which generates a bunch of detail to less-than-perfect footage. Currently, the free limit is three 10-second clips per week, so I used it on some old VHS camcorder footage.
I’m sorry, Grandpa.
Machine learning scientist Jochem Gietema found that the Image Playground app struggled to consistently identify his skin tone and hair texture, and exhibited racial biases when following prompt directions. This isn’t uncommon for AI image generation models, but it’s a blunder that Apple missed despite limiting Image Playground to only faces in illustrated styles, in part to avoid such behavior.
The Chinese AI app is no longer available on local app stores after acknowledging it had failed to meet Korea’s data protection laws. The web version is still accessible, and the app will return if and when it complies with the rules.
Italy blocked the app on similar grounds earlier this month, while the US and other countries are exploring bans for government and military devices.
I’m not talking about using something like Kagi, although that’s one way to do it. Last year, I directed Verge readers to a Tedium blog post explaining how to add “udm=14” to your browser’s default search URL to get rid of Google Search AI Overviews.
If that’s a headache, there’s also Tedium author Ernie Smith’s website that builds the “disenshittification Konami code” into your searches. Alternatively, you could just use cussword-laden searches.
[udm14.com]
Apple is still working on some big AI-powered upgrades to Siri, but instead of launching with iOS 18.4, they may be limited or delayed until iOS 18.5, Bloomberg reports.
The AI search engine’s new tool “performs dozens of searches, reads hundreds of sources, and reasons through the material to autonomously deliver a comprehensive report.” It’s similar to the research tools offered by ChatGPT and Google Gemini, but it’s available to free users with a “limited number of answers” each day.
I tried out Syntilay’s AI designed, 3D-printed footwear, which had me scan my feet with my phone for custom sizing. Syntilay says it wants to be “the content creator shoe company” by quickly designing shoes for YouTuber’s merch stores, but it’s just using consumer software like Midjourney to design these. Can’t anyone do that?
Free users can upload files to be analyzed, a feature previously limited to the paid tier. Reddit users report the feature is rolling out gradually. Gemini accepts various text, spreadsheet, and code file formats, or Android and web users can import files directly from Google Drive. You can upload multiple files simultaneously, though you can’t share an image and file in the same prompt.
ChatGPT has had file uploads for free users since May 2024.
[androidpolice.com]
Apple Intelligence is to be powered primarily by Alibaba, with a few features — including Visual Intelligence — handled by Baidu, according to Bloomberg. Apple’s own software will handle on-device processing, with Alibaba and Baidu stepping in for server-side tasks, including censorship.
Apple is reportedly targeting a mid-year launch, but has regulatory hurdles to clear first.
The Take It Down Act, co-sponsored by Sens. Ted Cruz (R-TX) and Amy Klobuchar (D-MN), would make it a crime to publish non-consensual intimate images, even if they’re AI-generated deepfakes. It would also force social media sites to have processes quickly remove these images when reported. The bill passed the Senate last year, too, but now has another chance to be taken up in the House.
The Wall Street Journal reports that The Verge’s parent company, Vox Media, and other publishers like Conde Nast, Forbes Media, and Politico filed a copyright and trademark infringement lawsuit (pdf) against the enterprise AI company Cohere. They say evidence shows Cohere uses unlicensed copies of content to directly compete with publishers, and they list 4,000 specific examples of “verbatim regurgitations and substitutional summaries of news content.”
On the Decoder podcast, we recently discussed similar media lawsuits against AI firms and spoke to Cohere CEO Aidan Gomez last summer.
[wsj.com]
The company completed its acquisition of meal-planning software startup Intent and plans to launch personalized meal features to fit users’ dietary needs, make nutrition logging easier, automate grocery lists, and even add ingredients to online store carts.
The Google AI assistant’s Deep Research mode now works on both Android and iPhone, though is still exclusive to paying Gemini Advanced subscribers. The mode, which can produce relatively detailed research reports, launched as a web-only option in December. It expanded to Android last week and iOS yesterday.
ChatGPT has its own deep research feature, which we tested last week to mixed results.
[9to5google.com]
Elon Musk offered to buy OpenAI for $97.4 billion on Monday, but according to Wired:
[Sam] Altman has also told employees that OpenAI’s board, which he sits on, has yet to receive an official offer from Musk and the other investors. If and when this happens, the board plans to reject the bid, according to those same sources. Internally, OpenAI employees reacted to the news with a mixture of fear and exasperation.
The Federal Trade Commission has finalized an order imposing the fine, which DoNotPay agreed to last year. In addition, the company is forbidden from deceptively advertising, without proof, that its “so-called robot lawyer” is as good as a human lawyer.
The agency approved the order by 5–0 on January 16th, prior to former FTC Chair Lina Khan’s exit and replacement with Andrew Ferguson.
