I Have a Warning for Everyone. Consider this my open admission. When I graduated from UC-Berkeley with my "useless" comparative literature degree, into one of the bleakest job markets in recent American memory, I thought to myself, . That was what brought me to marketing myself as an "academic editor," and an "admissions essay advisor," on various freelancing websites last fall. I figured I had done my fair share of editing for friends throughout the years, and I needed another gig to supplement my inconsistent substitute-teaching paychecks.

The company is reportedly making a unified app to streamline the user experience. OpenAI is developing a "super app" for desktop that unifies ChatGPT, its browser and its Codex app, according to the and . A company spokesperson told the publications that OpenAI Chief of Applications Fidji Simo will lead the application revamp with assistance from OpenAI President Greg Brockman. Simo will also help the marketing team advertise the app when it comes out. OpenAI's leadership is apparently hoping that combining several products can help it streamline user experience and dedicate its resources to one project.

Its march toward automation continues. Amazon has acquired Rivr, a startup focused on autonomous robotics. Rivr is based in Zurich and was valued at $110 million in a funding round from August 2024, which both Amazon and its CEO's Bezos Expeditions participated in. Financial details of the acquisition were not disclosed. Rivr's robots have four legs and wheels that allow it to maneuver on stairs and other potentially uneven surfaces.

A feature called Desktop Intelligence would let the AI pull context from open apps and your desktop. Google is testing a version of its Gemini app for macOS, reports . The app would bring the AI assistant to uncharted territory, and in more direct competition with OpenAI's ChatGPT and Anthropic's Claude, both of which offer standalone Mac apps. Gemini remains accessible through the web, and it sounds like the macOS app offers the same set of features, with the ability to respond to prompts, search the web and generate text, images and code. The major differentiator of the Mac app could be a feature called Desktop Intelligence, which gives Gemini a new source of information and context for its responses.

The company says humans will play a key role in important decisions. A little more than a year after ditching third-party fact checkers and rolling back much of its proactive content moderation, the company says it will further transform its approach by drastically reducing the number of human moderators in favor of AI-based systems. The company says the change will happen over the next few years, and will allow the company to catch more issues faster than its current approach. Meta didn't say how much of its contract workforce might be cut as it makes this transition. The company employs thousands of contractors around the world to review content flagged by its AI systems and user reports among other tasks.

Amazon's next-generation voice assistant launches in early access in Europe for the first time. Amazon's next-generation smart assistant has entered its Early Access program in the UK, marking Alexa+'s European debut following rollouts in the US, Canada and Mexico. Starting March 19, invitations to start using the smarter, more conversational will be sent out to hundreds of thousands of willing participants, Amazon said in a, adding that Alexa is the most popular voice assistant in the UK. As well as its more natural communication, agentic capabilities, contextual awareness and ability to remember previous conversations across devices, Amazon that users across the pond are getting an authentically British AI-powered assistant. It understands slang terms like cuppa and might even accuse you of taking the mick in the middle of a conversation.

The company had originally planned to announce global availability. A few hours after this story was published, Google reached out to retract the news. Given that the company says it is testing in more markets, it seems entirely possible that the global Search Live release will happen sooner than later. After rolling out Search Live to all US Google app users last September, Google is now bringing the feature to every place where it offers its AI Mode chatbot. Search Live, if you need a reminder, allows you to point your phone's camera at an object or scene and ask questions about what you see in front of you.

Sir Paul McCartney was among the artists who spoke out on the issue. After significant backlash, the UK backed off from that position. We have listened, Technology Secretary Liz Kendall said on Wednesday. However, the government's new stance is, well, not a stance at all. It currently no longer has a preferred option about how to handle the issue.

The tool is now available in every place where the company offers its AI Mode chatbot. After debuting in the US, Search Live is now available globally. After rolling out Search Live to all US Google app users last September, Google is now bringing the feature to every place where it offers its AI Mode chatbot. Search Live, if you need a reminder, allows you to point your phone's camera at an object or scene and ask questions about what you see in front of you. Google debuted the tool at I/O 2025 before it began rolling it out to users.

Safety is a desirable property that can immensely increase the applicability of learning algorithms in real-world decision-making problems. It is much easier for a company to deploy an algorithm that is safe, i.e., guaranteed to perform at least as well as a baseline. In this paper, we study the issue of safety in contextual linear bandits that have application in many different fields including personalized ad recommendation in online marketing. We formulate a notion of safety for this class of algorithms. We develop a safe contextual linear bandit algorithm, called conservative linear UCB (CLUCB), that simultaneously minimizes its regret and satisfies the safety constraint, i.e., maintains its performance above a fixed percentage of the performance of a baseline strategy, uniformly over time. We prove an upper-bound on the regret of CLUCB and show that it can be decomposed into two terms: 1) an upper-bound for the regret of the standard linear UCB algorithm that grows with the time horizon and 2) a constant term that accounts for the loss of being conservative in order to satisfy the safety constraint. We empirically show that our algorithm is safe and validate our theoretical analysis.