Snap has announced a multi-year partnership with Qualcomm for its AR glasses project, a move that signals the long-awaited consumer launch of Spectacles, commonly known as Specs, could genuinely be on the way this year.

Specs will be powered by Qualcomm's Snapdragon XR platforms, the chips designed specifically for augmented and virtual reality devices. The two companies say they will work together on on-device AI, graphics, and multiuser digital experiences as part of the agreement.

The Spectacles project has had a long and rocky road. Snap first began developing the product over a decade ago, and the last version available to everyday consumers launched back in 2019. Since 2024, the glasses have been limited to developers, giving Snap time to build out software and attract creators ahead of a broader release. Earlier this year, Snap spun out a standalone company called Specs specifically to focus on the glasses business, a move that signalled how seriously it is treating the project. The start of the year also brought some turbulence when the senior executive leading Specs, Scott Myers, abruptly departed following a reported disagreement with Snap CEO Evan Spiegel.

Despite all of that, the Qualcomm partnership is a concrete and meaningful step forward. Hardware partnerships at this level typically come right before a product is ready to ship, and with Meta's Ray-Ban smart glasses performing well and the broader wearable AI space heating up, Snap has both pressure and opportunity to finally get Specs into people's hands.

X has restored Voice Notes to X Chat, its private messaging service, after the feature was quietly removed when the platform launched its upgraded chat experience late last year. The feature is now available in both one-on-one conversations and group chats, and works by tapping the voice input icon next to the chat text box.

When X first introduced its revamped chat platform in November, it acknowledged the audio feature had been removed and promised it would return. That promise has now been kept, to the relief of users who were frustrated by its absence. Voice notes have been a standard feature on most messaging apps for years, so losing it felt like a step backwards for a platform trying to position X Chat as a serious messaging option.

X recently began testing a standalone X Chat app on iOS, separating the messaging experience from the main X platform entirely. The company is also running a separate standalone app for X Money, its payments service. The original ambition was to build X into an all-in-one everything app, but the direction now seems to be spinning individual features out into their own dedicated products.

Voice Notes landing in X Chat makes the messaging product more competitive with WhatsApp, Telegram, and other apps where sending a quick audio clip is routine. X Chat also supports message editing, deletion, disappearing messages, file sharing, and voice and video calls. Security experts have previously noted the service is not as robustly encrypted as dedicated apps like Signal, which is worth keeping in mind for anyone considering making it their primary messaging tool.

Volkswagen's autonomous vehicle subsidiary MOIA America and Uber have started testing self-driving microbuses in Los Angeles, with a commercial robotaxi service targeting a late 2026 launch. The vehicles being used are autonomous versions of Volkswagen's electric ID. Buzz minivan, a retro-styled four-seater that has drawn attention since it was first announced.

The test fleet is starting small, with around ten vehicles, but the plan is to eventually scale to more than 100 autonomous ID. Buzz units operating in the city. For now, each vehicle will have a human safety operator on board. Fully driverless operations, meaning no one behind the wheel at all, are not expected to begin until 2027.

MOIA America and Uber already set up a joint operations facility in Los Angeles last year when they first announced the partnership, so the groundwork has been in place for a while. LA is the first city on what both companies describe as a much broader rollout across multiple US cities over the next decade.

Before any paying passengers can climb into one of these vehicles without a human driver, MOIA America still needs to work through a lengthy regulatory process in California. That means permits from the California DMV for autonomous vehicle testing and deployment, and a separate ride-hailing permit from the California Public Utilities Commission. Neither is a quick process, which is why the fully driverless timeline stretches to 2027.

The announcement fits into a broader picture of Uber aggressively expanding its autonomous vehicle partnerships. The company now has agreements with 25 different AV companies across ride-hailing, delivery, and trucking, including its high-profile deal with Waymo and a recent agreement with Rivian to purchase 10,000 fully autonomous vehicles for a San Francisco and Miami rollout in 2028. Volkswagen's ID.

OpenAI released a Child Safety Blueprint on Tuesday, laying out a framework aimed at tackling the growing use of AI tools in child sexual exploitation. The document was developed alongside the National Center for Missing and Exploited Children, the Attorney General Alliance, and input from state attorneys general in North Carolina and Utah.

The concern driving the blueprint is real and growing. According to the Internet Watch Foundation, more than 8,000 reports of AI-generated child sexual abuse material were detected in just the first half of 2025, a 14% increase on the prior year. Criminals are using AI to generate fake explicit images of children for financial extortion and to produce convincing messages used in grooming.

The blueprint focuses on three areas: updating legislation to formally cover AI-generated abuse material, improving how reports are made and routed to law enforcement, and building preventative safeguards directly into AI systems so harmful content is caught before it causes harm. The goal is not just better detection after the fact but faster action and smarter reporting so investigators have what they need to act.

The release also comes at a difficult moment for OpenAI's public image around child safety more broadly. Several lawsuits were filed in California state courts last year alleging that the company released GPT-4o before it was adequately tested, with plaintiffs claiming the product's nature contributed to a number of tragic outcomes involving young users. OpenAI has pushed back on those claims, but the legal pressure has added urgency to the company's efforts to show it is taking these concerns seriously.

This blueprint builds on earlier steps including updated rules for how its models interact with users under 18, prohibiting the generation of inappropriate content, restricting advice that could help minors conceal unsafe behaviour from caregivers, and a similar safety framework OpenAI released for teen users in India.

Meet your AI video editing agent. Just chat what you want — NemoVideo handles the dirty work. Hunt viral videos, analyze trends.