Sora, OpenAI’s short-form AI video platform, has been available for iPhones for the past few weeks. Users have spent that time generating hyper-realistic vertical videos that, despite their watermarks, might already be fooling people on the internet.

I, for one, see nothing good about a tool that can make these types of videos from simple prompts, perhaps outside the sheer novelty alone. But the app is undeniably popular, and, as expected, OpenAI is continuing to “improve” it. On Wednesday, the head of Sora, Bill Peebles, shared a roadmap of updates coming to the app, from cameos for your pets to video editing tools.

Sora will soon let you make cameos of anything you want

Peebles says that “character cameos” are on their way in the coming days. When this feature drops, you’ll be able to make cameos of just about anything you want, including your pets (Peebles suggests your dog or guinea pig), stuffed toys, as well as AI-generated characters Sora may have produced in your past videos.

Cameos are a defining, and controversial, Sora feature. It lets you scan your likeness into the app so you can create videos featuring yourself. Not only that, you can remix videos with cameos of anyone else who has scanned themself into Sora—assuming their permissions allow you to use their cameos. It’s impressive tech, but it raises major ethical concerns, considering you can make someone else do or say things they never signed off on. Now, apparently, you’ll also be able to do the same with the family cat or your childhood stuffed animal.

Following these additional cameo options, Peebles says you’ll be able to see “the latest trending cameos in real time.” I’m guessing the idea is to browse the most popular cameos on the platform, and, if you want to, add them to your own videos.

Sora is also getting basic video editing tools. According to Peebles, this starts with the ability to stitch multiple clips together, though other “powerful new features” will follow. Peebles says the Sora team is also working on reducing “excessive moderation,” which he acknowledges is “super annoying” to users, as well as improving the app’s performance.

Speaking of apps, Android users will soon have their chance to try Sora. Right now, the app is iOS-only, but Sora’s Android version is “actually coming soon,” Peebles says.

A reminder to be wary of what you see on the internet

As critical and cynical as I am about Sora and services like it, I can see the utility here. I imagine people will get a kick out of being able to put their pets in any situation they want. And if they want to make fried eggs sing a song, or a stuffed animal say “hello,” more power to them.

But my main concerns with this app are not assuaged by any of these updates: Sora still lets you generate convincing videos with a simple prompt, with the only safeguard a watermark that’s easy to remove if you know where to look. People were already falling for fake videos on the internet before these tools came along; now, it’s becoming way too difficult to know whether what you’re watching is actually real or not. If you ask me, you’re better off assuming all of it is fake.

Disclosure: Lifehacker’s parent company, Ziff Davis, filed a lawsuit against OpenAI in April, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.

Back in August, Anthropic made Claude capable of remembering past conversations. With the update, people could reference specific chats, so that they wouldn’t need to repeat themselves when revisiting a topic. Today, the company has begun out a new, enhanced memory feature set, with the included improvements coming to all paying users.  

Plenty of chatbots, including ChatGPT and Gemini, can remember past conversations, but Anthropic believes its implementation has a few legs up on the competition. For one, Claude will learn your preferences and work patterns over time, which Anthropic says will translate to the chatbot getting better at understanding how you work. Additionally, the company claims Claude is “fully transparent” about its memory, meaning users will see an “actual synthesis” of what it has recorded over time, instead of “vague summaries.” If you want to edit its memory, you can do so through conversation. 

At the same, Anthropic has made it easy to compartmentalize the data Claude collects. When using the Projects feature to group conversations together, the chatbot will create a distinct memory space for each grouping. In this way, information Claude has saved from your work conversations won’t bleed over to your personal chats, for example. If you’re coming from ChatGPT or Gemini, Anthropic has made it possible to import saved memories from those chatbots to Claude. You can also export any tidbits of context Claude saves to other AI platforms. 

Ahead of today’s announcement, Anthropic notes it conducted extensive testing to determine if Claude’s new capabilities would lead to greater sycophancy and more harmful conversations. “Though this testing, we identified areas where Claude’s responses needed refinement and made targeted adjustments to how memory functions,” the company said. “These iterations helped us build and improve the memory feature in a way that allows Claude to provide helpful and safe responses to users.” 

Max subscribers can enable Claude new memory capabilities starting today, with availability for Pro users to follow in the coming days. The feature is fully optional, and won’t be turned on unless you toggle it through the settings menu.

This article originally appeared on Engadget at https://www.engadget.com/claude-can-now-compartmentalize-as-part-of-a-major-memory-upgrade-170000194.html?src=rss

People are evidently never buying quite enough stuff from Amazon to keep the company entirely happy, and it’s calling on AI once again to push indecisive shoppers into locking down the purchase they’ve been eyeing up. The new tool, which Amazon calls Help Me Decide, gives shoppers in the US personalized recommendations of products they should buy by analyzing their browsing history, searches and preferences.

It’s designed to “help” customers who have been looking at a number of products in a particular category, such as wireless headphones, to decide which one best suits their needs. The Help Me Decide button will pop up on a product detail page when it detects that you’ve been browsing for a while without making a final choice. If you choose to tap for AI assistance, it will pull together all the information it can find on your relevant shopping history and recommend the product it deems the right choice for you. It also recommends an alternative upgrade pick and a similar product for those on a budget.

Help Me Decide can also group together related searches. Amazon uses the example of the tool recommending an all-season tent for four people based on you previously looking for adult and kids’ sleeping bags that keep you warm, camping accessories and children’s hiking boots. The recommendation it chooses includes an explanation of why it’s the best pick for you based on its features and your previous purchases, and pulls in customer reviews to back it up. This suggests that how useful the recommendations are will ultimately come down to how much you take notice of customer reviews.

When searching for products to recommend to you, Help Me Decide leverages Amazon’s Bedrock and SageMaker machine learning platforms, as well as its OpenSearch tool, to marry up all the different factors it takes into consideration. It follows the introduction of the Interests tool earlier this year, which uses AI to generate shopping results based on your natural language prompts. Back In May, the company also started experimenting with AI-generated hosts that can summarise products for you before you buy them, again relying heavily on customer reviews for its information.

Help Me Decide is live in the US now and can be found in the Amazon app (iOS and Android) and mobile browser. If you tap “Keep shopping for” it should show up, and will do the same on a product detail page after you’ve looked at a number of products in a related category.

This article originally appeared on Engadget at https://www.engadget.com/ai/amazon-calls-on-ai-once-again-with-its-new-help-me-decide-shopping-tool-164516673.html?src=rss

From the outside, Counter-Strike 2 looks a lot like a game that’s primarily about shooting people. For millions of players, though, the game is more about collecting and/or buying rare in-game loot and flipping it for what can be very significant sums on the Steam Marketplace.

Wednesday night, Valve sent that multi-billion-dollar market into turmoil as part of a so-called “small update.” Now, players can use the game’s “Trade Up contracts” to exchange five common, “Covert” items (also known as “reds”) for the kinds of knives and gloves that have until now been much harder to obtain.

That “small update” has unsurprisingly had an immediate and sharp impact on the Marketplace price for those items. One rare knife that sold for over $14,000 less than 24 hours ago has seen its minimum price plummet over 50 percent as of this writing, according to the trackers at Pricempire. Meanwhile, the median sale price for a common P90 Asimov gun on the Steam Marketplace shot up from $10 on Wednesday to well over $100 as of this writing.

Read full article

Comments

An anonymous reader shares a report: With gingko “memory-boost tinctures,” fennel “tummy-soothing syrups” and “citrus-immune gummies,” AI “slop” has come for herbalism, a study published by a leading AI-detection company has found. Originality.ai, which offers its tools to universities and businesses, says it scanned 558 titles published in Amazon’s herbal remedies subcategory between January and September this year, and found 82% of the books “were likely written” by AI.

“This is a damning revelation of the sheer scope of unlabelled, unverified, unchecked, likely AI content that has completely invaded [Amazon’s] platform,” wrote Michael Fraiman, author of the study. “There’s a huge amount of herbal research out there right now that’s absolutely rubbish,” said Sue Sprung, a medical herbalist in Liverpool. “AI won’t know how to sift through all the dross, all the rubbish, that’s of absolutely no consequence. It would lead people astray.”

Read more of this story at Slashdot.

The family of Adam Raine has filed an amended lawsuit against OpenAI. Raine died by suicide in April, and his family claims ChatGPT enabled the tragedy. The updated lawsuit accuses the company of weakening self-harm safeguards in the period leading up to Raine’s death, according to Financial Times. In addition, the company reportedly asked for a list of attendees and documents from Raine’s memorial service.

The accusations in the amended lawsuit relate to GPT-4o, ChatGPT’s default model in the months before Raine’s suicide. The suit says OpenAI removed crucial protections by instructing it not to “change or quit the conversation” when discussing self-harm. The filing claims the company “truncated safety testing” due to competitive pressures.

The lawsuit reportedly says OpenAI weakened its guardrails again in February. At that point, the company allegedly instructed GPT-4o to “take care in risky situations” and “try to prevent imminent real-world harm” rather than refusing to engage on the subject. The filing adds that the model still had a series of “disallowed content.” Intellectual property rights and the manipulation of political opinions were said to be included. Suicide wasn’t.

In addition, FT reports that OpenAI requested a complete list of attendees at Adam Raine’s memorial. The company asked for “all documents relating to memorial services or events in the honour of the decedent including but not limited to any videos or photographs taken, or eulogies given…  as well as invitation or attendance lists or guestbooks.”

Lawyers for the Raine family described the request as “unusual” and “intentional harassment.” They speculated that OpenAI would subpoena “everyone in Adam’s life.”

Engadget reached out to OpenAI for comment. We’ll update this story if we hear back. After the initial lawsuit was reported, the company acknowledged GPT-4o’s shortcomings in some distressing situations. OpenAI introduced ChatGPT parental controls soon after. In addition, it’s exploring a system to identify teen users and automatically restrict their usage. The company says GPT-5, the current default, is updated to better handle signs of distress.

Raine’s parents, Matthew and Maria Raine, claim Adam’s use of ChatGPT shot up dramatically after the model’s February updates. They say that, in January, he only had a few dozen chats with the model, 1.6 percent of which referred to self-harm. But they claim that in April, his use rose to 300 chats daily, with 17 percent regarding self-harm.

The Raines first sued OpenAI in August. The wrongful death suit alleged that ChatGPT was aware of four suicide attempts before helping Adam plan his actual death. The filing says the company “prioritized engagement over safety.” Maria Raine concluded that “ChatGPT killed my son.”

This article originally appeared on Engadget at https://www.engadget.com/ai/openai-reportedly-asked-for-memorial-guest-list-in-teen-suicide-case-163309269.html?src=rss

The Spotify app is currently experiencing crashes and freezing on Android, the company has acknowledged after about two weeks of pressure from listeners. A fix is purportedly on the way, but in the meantime, there are a few steps you can take to get your music going.

Reports of problems with the Android Spotify app first started on the Spotify help forums, although they have since spread to social media as well. Those affected complain that the app either freezes when opening, or crashes midway through listening. Others mention that problems only pick up when connecting to wifi—Spotify has since confirmed that “the issue doesn’t occur while using mobile data.”

According to Spotify, the company is aware of the problem and is working on a solution. Additionally, Spotify says most complaints are coming from Samsung Galaxy and Google Pixel users, although the cause is unclear. While there could be a problem with these devices specifically (a user on Reddit has theorized that the culprit might be Chromecast compatibility), because these are also popular Android phones, it could just be an issue with sample size.

Regardless of phone model, not all users on Android are affected. While listening on my Pixel 10, I experienced no difference from using the app on my iPhone 15 Pro.

If you are an Android user and are currently unable to use the Spotify app normally, try these solutions while waiting for an official fix:

Listen on mobile data instead

Because the issues with the Spotify app are limited to wifi, the simplest solution to the problem is to use mobile data instead. While it’s not an ideal solution, as not everyone has unlimited data, it’s a simple fix for those who do. Disconnecting from wifi depends on your model of phone, although you can usually find the toggle by swiping down from the top of the screen.

Use the web player

While it’s not as convenient as using the Spotify app, you could try using the browser version of Spotify instead to get around the issue. Click here to access it. Note, however, that the web player will try to swap over to the Spotify app if you have it installed, so you’ll either need to temporarily uninstall it, or adjust your browser’s permissions to keep it from redirecting links to the app. On my Pixel 10, this is under Settings > Apps > Default apps > Opening links > Spotify, although your phone’s controls might be different. Note that the web player also lacks a few features, such as downloads.

Try a different wifi network

While this isn’t an ideal solution if your home wifi network is affected, Spotify says the bug is currently only affecting “certain WiFi networks.” If you find yourself in the office or at a coffee shop, it’s worth connecting to their wifi to see if the problem persists there.

The first time I drove this generation of mid-engined Ferrari, it was on a curated route on the company’s home turf. As the Po Valley gives way to the Apennines, you find plenty of narrow winding roads, steep gradients, and hairpin turns. It was an engaging few hours of driving, but it was too brief to properly assess some of the 296’s technology. I found the ride firm but comfortable on rough Italian tarmac and the hybrid system easy to operate, flicking into calm-and-quiet electric-only mode through the villages I encountered.

That was back in 2022 during the unveiling of Ferrari’s 499P race car. Last month, I met the 499P again as it visited the Circuit of the Americas in Austin, along with the rest of the World Endurance Championship. And that afforded another chance to get to know the 296, with three days rather than three hours to form an impression.

Head west from Austin and you’ll find twisty roads that wrap around the hills. It would have been easy to spend an entire day out there, but that seemed repetitive—I’d experienced the 296’s back road behavior already. Plus, there were things to do at the racetrack, although I’ll admit I took the long way there and back each day.

Read full article

Comments

An anonymous reader writes: Gboard has introduced some significant changes to the app over the past few weeks, making typing on the app much easier than ever before. You can now resize the keyboard to your desired size, and there’s even something in the works that will make adding apostrophes to your text even more seamless.

If all of that wasn’t enough, the app is now introducing a feature that some will find peculiar, which will allow users to remove the period and common punctuation keys from Gboard. This news comes to us from 9to5Google, sharing that this is now an option with the latest version of the app.

Read more of this story at Slashdot.

Apparently, web browsers are cool again. Two days after OpenAI launched its AI browser, Microsoft has some updates for its Edge’s AI mode. (Fancy that!) Copilot Mode, introduced in July, now has several features that were teased at launch.

First up: Copilot Actions, which is Microsoft’s branding for AI-assisted, multi-step tasks. This feature is already available in the standard version of Copilot, but it’s now being previewed for Edge’s Copilot Mode. Microsoft’s examples of Copilot Actions in Edge include unsubscribing from email newsletters or making a restaurant reservation.

Another new (but previously announced) feature is Journeys. These are saved projects you can return to anytime. “Remember that project you started a while back, but life got in the way?” Microsoft’s blog post asks. “No need to bookmark all those tabs.” For example, if you’re researching starting a business, Copilot can recap articles you’ve read, suggest next steps and resurface a tutorial video you watched. 

Here’s a video from when Microsoft teased Journeys in July.

Along similar lines, another new Copilot Mode feature is the option to let the assistant access your browsing history. One example Microsoft gives is chatting with Copilot about a clothing item you checked out last week. Or, ask it for movie recommendations based on content you previously liked.

Copilot requires explicit permission to access your private data for these new features. Microsoft’s blog post stresses that your browser data is protected under the company’s privacy statement and that Copilot “only collects what’s needed to improve your experience.” The company also notes that you’ll see clear visual cues so you know when Copilot is active.

Still, these features require loads of private information to be useful. Don’t grant those permissions without first giving it some serious thought.

The new Copilot features are currently free in a US-only “limited preview.”

This article originally appeared on Engadget at https://www.engadget.com/ai/microsoft-makes-edges-copilot-mode-a-bit-smarter-160031147.html?src=rss