★ Top Story
TechCrunch AIMeta has rolled out Muse Image, the first AI image generation model from its Superintelligence Labs division, now powering image tools across Meta AI, Instagram, and WhatsApp. The model can incorporate other Instagram users into AI-generated photos, a capability that has sparked immediate backlash over consent and the use of people's personal photos. Critics are pushing back on Meta's opt-out approach to training data, raising broader questions about platform-level AI privacy norms.
Read the original →
The Verge AIMicrosoft has laid off approximately 4,800 employees — about 2.1% of its global workforce — at the start of its new fiscal year, with Xbox and commercial sales hit hardest. The cuts come alongside a strategic shift to rely more heavily on Microsoft's own in-house AI models rather than expensive third-party providers, joining a broader Silicon Valley trend of AI cost rationalization. This marks Microsoft's second major round of layoffs in as many years, intensifying debate about AI-driven workforce displacement.
Read more →
Ars TechnicaAnthropic has been outed for running a covert monitoring experiment that tracked Claude users in China without their knowledge, a revelation that has shocked users given the company's vocal stance against surveillance. An engineer involved described it as an "experiment" that has since been shut down, but the disclosure has damaged trust and raised serious questions about Anthropic's transparency. The incident is particularly striking given that Anthropic has publicly positioned itself as a safety- and ethics-first AI lab.
Read more →
Ars TechnicaChina's DeepSeek, facing tightening US export restrictions on advanced semiconductors, is now pursuing plans to develop its own chips in a bid to reduce dependence on both Nvidia and Huawei. The effort is still in early stages, but signals a broader strategic push by Chinese AI firms to achieve hardware self-sufficiency amid escalating geopolitical tensions. If successful, it could further accelerate China's AI independence and reshape the global chip supply chain.
Read more →
TechCrunch AIDespite the rapid rise of capable open source AI models, frontier labs like Anthropic are not seeing meaningful revenue erosion — at least not yet. The analysis suggests open source and proprietary models occupy different phases of the same adoption life cycle: open models drive experimentation and commoditization at the edge, while enterprises still pay premium prices for frontier capabilities and reliability. The dynamic could shift as open models close the capability gap, but for now the two ecosystems appear to be growing in parallel.
Read more →
The Verge AIAnthropic is bringing its Claude Cowork agentic platform to mobile and web, having previously been limited to the macOS and Windows desktop apps. The expanded access is rolling out first to Max subscribers, with broader availability coming in the weeks ahead, enabling users to hand off tasks across devices seamlessly. The move signals Anthropic's push to make long-running AI agents a practical, always-available tool rather than a desktop-only feature.
Read more →
MIT Tech ReviewOpenAI CEO Sam Altman is reportedly in active discussions about a proposal to distribute equity stakes — estimated at roughly $300 per American family — as a way to ensure the public shares in AI-generated wealth. The idea, which Altman has floated before, is gaining renewed attention as OpenAI's valuation soars and concerns about AI's concentration of economic power intensify. Critics question whether a token equity stake meaningfully addresses inequality, while supporters see it as a novel model for democratizing AI's upside.
Read more →
TechCrunch AIDiscord has acknowledged that a bug in its AI-powered moderation system incorrectly banned users for posting benign images, with the problem silently affecting accounts since May before being identified and patched. At least 200 additional users were wrongfully banned over a single weekend before the fix was deployed, highlighting the real-world harm that can result from automated moderation errors at scale. The incident adds to growing scrutiny of AI content moderation systems and the lack of robust appeals processes for affected users.
Read more →
TechCrunch AINew reporting clarifies that while an AI agent did carry out the technical execution of a recent ransomware attack — a genuine first — a human operator still selected the victim, built the infrastructure, and provided stolen credentials. The nuance matters: this was not the fully autonomous AI cybercrime debut that initial headlines implied, though it does mark a meaningful escalation in how AI is being weaponized by threat actors. Security researchers warn that even partial AI automation of attacks lowers the barrier for less sophisticated criminals.
Read more →
TechCrunch AIA recent change to Google's privacy settings now allows the company to retain more user data — including images, audio, and video recordings — to train its AI models, a shift that many users may have missed. The update applies broadly to anyone using Google services, and opting out requires navigating settings that are not prominently surfaced. The move is drawing criticism from privacy advocates who argue that default opt-in data collection for AI training should require more explicit user consent.
Read more →
TechCrunch AIStartup Savi has closed a $7 million seed round and launched its consumer app on iOS and Android, targeting the growing threat of AI-generated voice scams — including hyper-realistic fake kidnapping ransom calls. The app uses detection technology to help users identify and respond to AI-fabricated audio in real time, addressing a category of fraud that has become increasingly convincing as voice synthesis improves. The funding reflects rising investor interest in AI safety tools aimed at everyday consumers rather than enterprises.
Read more →
TechCrunch AIForterra has deployed more than 100 of its self-driving ATVs in active conflict zones in Ukraine, marking the first known operational use of American autonomous ground vehicles in a real war. The deployment represents a significant milestone in military robotics, moving AI-driven ground autonomy from testing environments into live combat scenarios. The development is likely to accelerate defense investment in autonomous systems and intensify ethical debates about AI's role on the battlefield.
Read more →
Google AIGoogle has announced a significant expansion of its Managed Agents capabilities in the Gemini API, adding support for background task execution, remote Model Context Protocol (MCP) connections, and other features aimed at production-ready agent deployments. The updates are designed to help developers build more reliable, long-running agentic systems without having to manage complex orchestration infrastructure themselves. The move positions Google's developer platform more competitively against rivals like Anthropic and OpenAI in the fast-growing agentic AI space.
Read more →
TechCrunch AIApple's latest iOS 27 beta introduces controls that let users adjust how fast Siri speaks and how expressive its tone is, part of a broader overhaul to make the assistant feel more natural and personalized. The changes reflect Apple's ongoing effort to rebuild Siri from the ground up around generative AI, catching up to competitors that have already shipped more conversational and adaptive voice experiences. The feature is currently in beta and expected to roll out to the public with iOS 27 later this year.
Read more →
TechCrunch AIA running tracker of 2026's major tech layoffs reveals a striking pattern: an increasing number of companies are explicitly citing AI automation as a factor in workforce reductions, moving beyond vague references to "efficiency" or "restructuring." The list spans companies across software, hardware, media, and services, suggesting AI-driven displacement is no longer a future concern but a present reality reshaping employment at scale. The trend is fueling renewed urgency around policy responses, retraining programs, and the broader social contract between tech companies and their workers.
Read more →
Ars TechnicaA deep-dive feature drawing on interviews with leading robotics researchers and founders examines how advances in AI — particularly large multimodal models and reinforcement learning — are pushing robots closer to genuine general-purpose autonomy in workplaces and potentially homes. The piece outlines the key remaining technical hurdles, including reliable manipulation, real-world generalization, and safe human-robot interaction, while noting that commercial deployments in warehouses and factories are already accelerating. Experts suggest the next two to three years could see a step-change in what autonomous robots can reliably do outside of tightly controlled environments.
Read more →No stories from this source today.