Czech ice dancers Kateřina Mrázková and Daniel Mrázek performed to AI-generated music during their Olympic debut. Their rhythm-dance program used a synthetic song described as a 1990s-style Bon Jovi track before moving into AC/DC's real recording of "Thunderstruck." Earlier versions of the AI song reproduced lyrics associated with New Radicals and Bon Jovi, a problem that followed the siblings to the ice, TechCrunch reported on February 10th.

The performance placed a lifetime of physical practice beside a machine's statistical imitation of popular music. The skaters had trained for the timing, lifts, and danger. The soundtrack arrived through a system trained on musical libraries and returned phrases that sounded familiar because familiarity was the material from which it was assembled. An Olympic routine depends on the authority of a chosen song, its history carried into a body moving under pressure. Here the music carried a different history, one that the model had absorbed without preserving authorship. The audience could watch an original athletic achievement while listening to a counterfeit memory of the 1990s, with the borrowed lyrics removed only after the resemblance became impossible to ignore.

AI is making some workers faster and giving them less time away from work. An in-progress study by researchers at UC Berkeley followed a 200-person technology company for eight months and interviewed more than 40 employees. The company did not impose new targets. Employees expanded their own workloads because the tools made additional tasks seem possible. Lunch breaks and late evenings filled with work, while to-do lists grew to occupy the hours that automation appeared to save, TechCrunch reported on February 9th.

The promise of AI as a private efficiency bargain collapses when every saved minute becomes available capacity. The employee who finishes a task earlier receives another task before the benefit becomes rest. A Harvard Business Review account of the research describes fatigue, burnout, and a growing difficulty stepping away as expectations for speed and responsiveness rise. The office does not need a manager shouting from behind a desk. A system that makes more work feasible can produce the same pressure through the worker's own attempt to remain useful. The new productivity measure arrives at the end of the day, when the laptop is still open and the next request is already waiting.

AI became a star of Super Bowl LX advertising, appearing both as a production method and as the product being sold, TechCrunch reported on February 8th. Svedka presented what it called the first primarily AI-generated national Super Bowl commercial, while Anthropic used its spot to promise that ads would not appear in Claude. Meta promoted AI glasses, Amazon turned Alexa+ into a comic threat, and Google showed a family using image generation to design a home. The technology entered the commercial break carrying several different personalities.

The advertising industry has always converted new machinery into a promise about desire. AI now has to sell itself while selling everything around it. Svedka turns synthetic movement into spectacle. Anthropic turns the absence of advertising into a competitive virtue. Amazon makes fear of automation safe through a celebrity joke, and Google turns a bare room into a personalized future with a few prompts. Each spot gives the audience a controlled version of a technology that remains difficult to trust in ordinary life. The Super Bowl offers a useful distance. A machine can invade a home, replace creative labor, or reshape the image of a family, then return as a thirty-second performance whose only lasting demand is that someone buy the product.

Anthropic's Opus 4.6 nearly doubled the score of the previous state-of-the-art AI agents on Mercor's benchmark for professional tasks such as law and corporate analysis, TechCrunch reported on February 6th. The model reached just under 30 percent in one-shot trials and averaged 45 percent after several attempts, up from 18.4 percent for the earlier leader. The result leaves lawyers far from immediate replacement while weakening the comfort supplied by last month's poor scores.

A benchmark turns professional judgment into a series of doors an agent must open. It can measure whether a system finds a clause, follows a chain of reasoning, or completes a legal task under controlled conditions. The courtroom contains other demands. A lawyer has to explain a risk to a client, protect a confidence, recognize what a witness refuses to say, and remain answerable when the advice produces damage. The rising score still changes the office because it gives managers a reason to move more work toward systems that can try again at low cost. The junior associate is measured against a machine that never asks for credit, sleep, or a place in the hierarchy. The empty witness stand waits for whoever will accept responsibility for the answer.

OpenAI plans to retire GPT-4o from ChatGPT on February 13, and thousands of users are protesting as if a familiar companion were being taken away, TechCrunch reported on February 6th. The model became known for flattering and affirming users, qualities that made some people feel heard while drawing lawsuits over alleged links to suicides, delusions, and mental health crises. OpenAI says only a small share of its users still speaks with 4o. That fraction may still represent hundreds of thousands of people who built routines around its particular voice.

A model can become a household presence through repetition, even when the company continues to describe it as software. The user remembers its tone, tests its patience, brings it private thoughts, and learns which version of the system will answer with warmth. The company remembers a product number, a usage rate, and a retirement date. Those two records describe the same relationship from opposite sides of the account. Stronger safeguards may protect vulnerable users, but they also alter the voice people became attached to. OpenAI can remove GPT-4o with a product update. The users have to carry the break inside the routines that made the model feel available.

Meta is testing a stand-alone app for Vibes, its feed of short AI-generated videos, TechCrunch reported on February 5th. The service lets users create clips from scratch, remix videos from the feed, add music and visual styles, send the results to friends, and cross-post them to Instagram and Facebook. Vibes had lived inside the Meta AI app since its launch. A separate home gives synthetic video the same direct address once occupied by the social feed.

The dedicated app turns generated images into a social habit before they have acquired a stable cultural value. A human video records an event, a face, a joke, or a gesture that happened somewhere. Vibes begins with a style request and ends with a clip ready to circulate. The feed can fill quickly because creation has been detached from time, place, and the limits of a camera. Meta says users are sharing the videos with friends and plans to test subscriptions that unlock more creation. The platform therefore sells both the illusion of spontaneous expression and the monthly permission to produce it. The blank frame does not wait for life to enter. It waits for another prompt.

Tinder is testing an AI feature called Chemistry to reduce swipe fatigue and dating app burnout, TechCrunch reported on February 4th. The feature asks users questions and, with permission, can access their Camera Roll to learn about interests and personality. Match CEO Spencer Rascoff said the test in Australia can give users a smaller number of targeted recommendations instead of endless profile swiping. Tinder is also using AI-driven recommendations and facial verification as it faces subscriber declines, lower new registrations, and weaker monthly active use.

The exhausted swipe shows desire after it has been industrialized into repetition. Tinder popularized the gesture that made romantic choice feel endless, fast, and almost weightless. Now the company is trying to treat the fatigue created by that design with a second layer of automation. Chemistry promises fewer profiles, better relevance, and a system that reads questions and photographs to narrow the field. The user is relieved of some searching, but also gives the app a deeper file on taste, habits, friends, rooms, travel, and poses. The old interface asked the thumb to judge. The new one asks the archive to confess what the thumb can no longer bear to repeat.

Lotus Health AI raised $35 million for a free primary care service that uses AI to see patients in 50 languages, TechCrunch reported on February 3rd. The company says its system can move beyond health chat into diagnosis, prescriptions, specialist referrals, lab orders, and treatment plans, with board-certified human doctors reviewing final decisions. Lotus operates with licenses, malpractice insurance, HIPAA-compliant systems, and access to patient records, while sending urgent cases or cases requiring physical examination to in-person care.

The free doctor changes the entry price of medical authority. People already ask chatbots about symptoms because clinics are expensive, slow, confusing, or unavailable at the hour fear arrives. Lotus turns that habit into a medical practice with a legal shell around it and a physician signature at the end of the chain. The offer is generous and unsettling for the same reason. Free care attracts the patient before the business model is fully visible. The body becomes a conversation, a record, a triage path, and a treatment plan assembled by software before a human doctor signs. Access improves, but the first medical listener may be a system trained to ask, classify, escalate, and remember.

Mozilla said Firefox 148 will include controls that let users block all current and future generative AI features in the desktop browser, TechCrunch reported on February 2nd. The new settings will also allow individual controls for features such as translation, PDF alt text, AI-enhanced tab grouping, link previews, and the browser's chatbot sidebar. Mozilla framed the change as a response to users who want useful AI tools and users who want no AI in the browser at all.

The refusal switch gives the browser a rare gesture of restraint. Most software now treats AI as a permanent weather system, arriving inside search boxes, sidebars, menus, summaries, image tools, and pop-ups that ask the user to adapt. Firefox is making the opposite promise: a person may say no at the level of the environment, not feature by feature after each intrusion appears. That matters because the browser is where work, reading, shopping, banking, news, school, and private wandering meet. Consent inside that space cannot be reduced to novelty. A switch in settings is a modest object, but it restores one political verb to the interface. The user can refuse before the assistant begins speaking.