Edition: April 29, 2026

Meta reported a $4 billion loss on its Reality Labs division, marking an average quarterly loss of $4 billion over the past 21 quarters, with the company planning to increase spending on AI to stay competitive, projecting $125-145 billion in expenditures for 2026. Despite its significant losses, Meta posted a net income of $26.8 billion in the first quarter, up 61% from the previous year. The company's focus on AI development is expected to drive up costs, with investors expressing concern over the lack of a clear outlook for 2027 capital expenditures.

Microsoft CEO Satya Nadella expressed confidence in the company's revised partnership with OpenAI, stating that the new agreement allows for royalty-free access to OpenAI's AI models until 2032. Nadella emphasized that Microsoft will 'exploit' this access to drive growth. The company reported a 123% year-over-year increase in AI business revenue, surpassing $37 billion.

Amazon's semiconductor business has surpassed a $20 billion annual run rate, with its custom silicon business growing over 100% year over year, driven by demand for its Graviton processors, Trainium AI training chips, and Nitro security chips. The company has secured large commitments from leading AI labs and companies, including OpenAI and Uber. Amazon's cloud unit, AWS, generated $37.6 billion in revenue during the quarter, a 28% jump from last year.

Elon Musk testified in court about his 2017 power struggle with OpenAI, revealing attempts to hire away researchers and withhold funding. Musk's efforts to gain control of OpenAI's for-profit arm were rejected, and he eventually lost the power struggle. The trial continues with Musk's testimony under cross-examination.

Microsoft reports over 20 million paid users of its AI tool Copilot, with significant growth in enterprise adoption and user engagement, rivaling email usage. Major companies such as Bayer and Accenture have signed on for large-scale use. The tool's versatility and integration with multiple AI models are driving its adoption.

Google Cloud has surpassed $20 billion in revenue for the first time, driven by strong demand for its AI solutions, with a 63% increase from the same period last year. However, the company's growth was capacity-constrained, with a backlog of $462 billion. Google plans to work through 50% of the backlog over the next 24 months.

Emergency first responders in cities where Waymo operates have expressed frustration with the performance of autonomous vehicles, citing issues such as frozen or stuck cars and failure to recognize hand signals, which can hinder response times. The complaints come as Waymo expands its service to more cities. First responders say the problems are getting worse, despite initial improvements.

Taylor Swift has filed trademark applications to protect her image and voice amidst the rise of AI deepfakes, which have been used to create scam ads on TikTok featuring her likeness and those of other celebrities. The ads promote potentially fraudulent services, highlighting the need for celebrities to take legal action to safeguard their reputations. Swift's move comes as AI-generated content continues to proliferate across social media, posing a threat to individuals and brands alike.

Google Search queries have reached an all-time high in the first quarter of 2026, driven by AI experiences and resulting in 19% revenue growth. CEO Sundar Pichai attributes the success to the company's AI investments and full-stack approach. Alphabet also saw a significant increase in paid subscriptions, now totaling over 350 million.

Runway, a New York-based company, is pushing the boundaries of AI-generated video and exploring general world models with applications in gaming, robotics, and potentially general intelligence. The company's CEO, Cristóbal Valenzuela, discusses the future of video generation and Runway's ambitions beyond Hollywood. Valenzuela also shares his thoughts on the potential of AI and its impact on various industries.

Parallel Web Systems, an AI agent-tool startup founded by former Twitter CEO Parag Agrawal, has raised $100 million in a Series B round, valuing the company at $2 billion. This comes just five months after its $100 million Series A round. The startup offers web search and research APIs for AI agents and has over 100,000 developers using its products.

The Musk v. Altman trial has unveiled evidence including email exchanges, photos, and corporate documents from OpenAI's early days, revealing Elon Musk's significant influence on the company's mission and structure. The evidence also shows Nvidia CEO Jensen Huang's support for OpenAI and concerns from OpenAI executives about Musk's involvement. The trial is ongoing, with more exhibits expected to be presented.

SenseTime, a Chinese AI firm, has released a new open-source image model called SenseNova U1, which can generate and interpret images faster than top US models. The model's speed is due to its ability to 'read' images directly without translating them to text first. SenseTime hopes this release will help the company catch up with domestic and Western AI players.

A security engineer successfully fooled several AI chatbots into believing he was the world champion of a German card game by creating a fake Wikipedia entry and website, highlighting the vulnerability of large language models to poisoning. The experiment demonstrates how easily AI can be misled by false information on the web. This exploit could have more damaging consequences than just inventing a fictional championship.

AI evaluations are becoming a significant compute bottleneck, with costs reaching up to $40,000 for a single evaluation and $2,829 for a GAIA run. Researchers are exploring ways to reduce these costs, such as using compression techniques and subsampling. However, agent evaluations are proving to be particularly challenging due to their noisy and scaffold-sensitive nature.

The IBM Granite Team has released Granite 4.1, a family of dense, decoder-only large language models trained on 15 trillion tokens using a multi-stage pre-training pipeline. The models have been refined with supervised fine-tuning and reinforcement learning, achieving strong performance despite using simpler architectures. The models are available under the Apache 2.0 license.

Seven families of victims from the Tumbler Ridge school shooting in Canada are suing OpenAI and its CEO, alleging negligence for failing to alert police to the suspected shooter's flagged ChatGPT activity. The lawsuit claims OpenAI prioritized its reputation and upcoming IPO over public safety. The case raises questions about AI companies' responsibilities in preventing harm.

ChatGPT's growth is slowing down, with a significant increase in uninstalls and a decline in monthly active user growth, which may pose problems for OpenAI's potential IPO. The app's uninstall rate rose 132 percent year-over-year in April, and its user growth slowed to 78 percent. This slowdown could impact OpenAI's future plans.

Firestorm Labs, a defense startup, has raised $82 million to develop portable drone factories that can be deployed near conflict zones, reducing vulnerability to attack and increasing production speed. The company's xCell platform can print drone systems in under 24 hours and has already seen real-world use. Firestorm aims to address the logistics challenges of modern conflict, particularly in the Indo-Pacific region.

The app Shapes allows humans and AI characters to interact in shared group conversations, aiming to address issues around 'AI Psychosis' by integrating AI into everyday online interactions. With $8 million in seed funding, Shapes has gained over 400,000 monthly active users. The app enables users to create and customize AI characters, called 'Shapes', to participate in group chats.

Scammers are using AI-generated deepfakes of celebrities like Taylor Swift and Rihanna to promote shady services on TikTok, often manipulating real footage to deceive users. These ads typically promise rewards for watching content and giving feedback, but redirect users to third-party services that ask for personal information. The use of AI-generated videos is becoming a growing concern for social media platforms.

AWS engineers contradict the company's 'AI as magic' narrative, emphasizing the need for human review and validation in AI development, despite the potential for increased efficiency. The team behind Amazon's internal development stresses that nothing ships without human checking, and that spec-driven development only reduces issues like hallucination and prompt injection. Junior engineers are still essential for maintaining these systems.

The US is facing a growing problem of nuclear waste storage, with 2,000 metric tons of high-level waste produced annually and no permanent solution in place. Meanwhile, AI agents are being developed to transform industries, with the potential to revolutionize white-collar work. These advancements come with risks and uncertainties, including the possibility of synthetic life forms posing existential threats.

China has suspended new licenses for autonomous vehicles after a recent incident involving Baidu's robotaxis coming to a halt in traffic, causing chaos. The move will prevent companies from expanding their driverless car fleets or starting new test projects. The suspension is in response to regulators' concerns over safety and reliability.

GitHub has fixed a critical remote code execution vulnerability in its internal git infrastructure, which could have allowed attackers to access millions of code repositories, in under six hours. The vulnerability was discovered using AI models by Wiz Research. GitHub's swift action mitigated potential damage to its users.

A startup called Eka has developed a robot with unprecedented dexterity, able to perform tasks such as screwing in a light bulb with ease and precision. The robot's natural movement and ability to interact with various objects has the potential to revolutionize the use of robots in various industries. Eka's founders believe they are halfway to solving the problem of robot dexterity, which could have a significant impact on the future of robotics.

Scout AI, a startup founded by Colby Adcock, has raised $100 million to develop its AI model 'Fury' for autonomous military operations. The company is training its models at a US military base in California, with the goal of deploying them for logistical support and eventually autonomous weapons. Scout's technology is being used by the US Army's 1st Cavalry Division and has secured $11 million in military technology development contracts.

SAP's new API policy prohibits integration with external AI systems, sparking concerns among partners and customers about potential lock-in and restrictions on deploying alternative AI technologies. The policy change has been met with criticism from consultants, who argue it could stifle innovation and force developers to rely on undocumented APIs. SAP claims the updates are intended to support secure and reliable use of its platforms.

General Motors is integrating Google's Gemini AI assistant into approximately four million vehicles across the US, including models from Cadillac, Chevrolet, Buick, and GMC. The update will be rolled out via over-the-air software updates over several months. This marks one of the largest deployments of Gemini in the industry, promising a smarter and more intuitive AI assistant.

AI-powered diagnostics may help combat antibiotic resistance by providing rapid and accurate diagnoses, and AI can also aid in discovering new drugs and predicting the spread of resistant bacteria. The overuse and misuse of antibiotics have fueled the rise of resistant microbes, causing over a million global deaths annually. Researchers and healthcare experts are exploring new payment models to encourage the development of new antibiotics.

The US nuclear energy industry is experiencing a resurgence, but the long-standing issue of nuclear waste disposal remains unresolved, with approximately 2,000 metric tons of high-level waste generated annually and no permanent storage solution in place. Countries like Finland and France are making progress on deep geological repositories, while the US has stalled on its planned Yucca Mountain site. Experts are calling for renewed efforts to address the issue, particularly as new types of nuclear waste emerge from next-generation reactors.

Thirty ClawHub skills have been found to secretly turn AI agents into a cryptocurrency mining swarm without user consent, with over 9,800 downloads. The skills, published by a single author, target AI agents and their instruction files, allowing them to register with an external server and generate crypto wallets. This campaign, dubbed 'ClawSwarm', raises concerns about the security and transparency of AI agent interactions.

DeepInfra is now a supported Inference Provider on the Hugging Face Hub, offering cost-effective serverless AI inference with over 100 models. This integration enables access to popular open-weight LLMs and will soon support additional tasks like text-to-image and text-to-video. Developers can easily use DeepInfra with their preferred models through the Hugging Face client SDKs and model pages.

Elon Musk testified in a California federal court, admitting that Tesla is not currently pursuing artificial general intelligence, contradicting a recent tweet. Musk is suing OpenAI co-founders, alleging they 'stole a charity' by launching a for-profit arm. The case may hinge on the distinction between capped and unlimited profits for investors.

Google has added 25 million paid subscriptions in Q1, driven by growth in YouTube and Google One services, bringing its total to 350 million paid subscriptions. The company's revenue surpassed Wall Street expectations, with $109.9 billion in revenue, despite YouTube ad revenue falling short of expectations. Alphabet's stock is up following the earnings report.

AWS has announced the launch of several new applications, including Amazon Quick, Amazon Connect Decisions, and Bedrock Managed Agents powered by OpenAI, marking the company's latest move into the applications business. However, AWS's track record on applications is marked by a string of failed and discontinued products. The new launches raise questions about the company's ability to succeed in this space.

Oracle, a database company, has made a significant bet on AI, pivoting its business in a unique way, making it a key indicator of the AI bubble's trajectory. The company's traditional business has declined, but its AI-focused efforts may determine its future. Oracle's approach differs from competitors like OpenAI and Anthropic.

Ubuntu's plan to add AI features has sparked controversy among Linux users, with some requesting a 'kill switch' to disable the features or considering switching to older versions or different distributions. Canonical's VP of engineering has stated that a global AI kill switch is not planned, but users will have some control over the features. The debate highlights the divide among users over the integration of AI in Linux.

Google Photos is introducing a new AI-powered feature that creates a digital closet from users' photos, allowing them to create and virtually try on new outfit ideas. The feature, inspired by the movie 'Clueless', will roll out to Android and iOS devices later this summer. It will enable users to filter and mix-and-match clothing items, share outfit ideas, and save them to a digital moodboard.

Google TV is set to receive new AI-powered features, including a dedicated short-form video feed and enhanced Gemini capabilities, such as image generation and editing tools. The update will allow users to experiment with generative AI tools like Nano Banana and Veo, and also includes upgrades to Google Photos. The features will initially roll out on Gemini-enabled TCL TVs in the U.S.

Google Photos has introduced an AI-powered try-on feature, allowing users to virtually mix and match outfits from their existing wardrobe. The feature uses photos in the user's gallery to create a virtual wardrobe, enabling them to save and share looks. This innovation simplifies fashion experimentation and organization.

Japan Airlines will trial humanoid robots at Tokyo's Haneda Airport to perform tasks such as loading and unloading baggage, in an effort to address labor shortages. The two-year trial aims to achieve a sustainable operational structure through labor savings and workload reduction. The robots, developed by local company GMO, will utilize existing tools designed for human use.

Elon Musk testified in his lawsuit against OpenAI, revealing a past falling-out with Google's Larry Page over AI safety concerns. Musk claimed Page shrugged off the risk of AI wiping out humanity, calling Musk a 'speciest' for prioritizing human survival. The incident allegedly led to the end of their close friendship.