★ Top Story
The VergeAnthropic has launched Claude Fable 5, the first model from its high-end Mythos class to be made available to the general public. The company says it leads competing models on software engineering, knowledge work, and vision tasks — with its advantage widening on longer, more complex jobs. The release comes with hard guardrails blocking responses in sensitive domains including cybersecurity, biology, and chemistry.
Read the original →
TechCrunchOpenAI has submitted a confidential IPO filing, joining rival Anthropic in racing toward a public market debut. The back-to-back filings from the two dominant AI labs signal a new era for the industry, with both companies potentially entering public markets within a similar window. The moves are expected to reshape the competitive and financial landscape of frontier AI development.
Read more →
TechCrunchApple's annual developer conference was dominated by AI, with the company unveiling a significantly upgraded "Siri AI" alongside iOS 27 and a broad set of new Apple Intelligence capabilities. Key announcements included AI-powered photo editing tools, an AI-enhanced Shortcuts app for building natural-language workflows, and new Safari and Password app features. Critics noted Apple was largely catching up to competitors, though its privacy-first framing — including on-device and confidential cloud processing — remained a distinctive selling point.
Read more →
Ars TechnicaApple is expanding its Private Cloud Compute infrastructure to Google Cloud, but insists user data remains inaccessible to Google throughout the process. NVIDIA GPUs with Confidential Computing technology are being used to enable secure, confidential inference for Apple Foundation Models running outside Apple's own data centers. The arrangement, announced at WWDC, is designed to let Apple scale its AI capacity without compromising its core privacy guarantees.
Read more →
NVIDIA BlogNVIDIA has confirmed that its Confidential Computing-enabled GPUs are now part of Apple's Private Cloud Compute stack, supporting server-side inference as Apple scales beyond its own data centers. The technology ensures that even cloud infrastructure operators cannot access user data during AI processing. The partnership, revealed at WWDC, represents a significant expansion of confidential AI computing at consumer scale.
Read more →
Ars TechnicaGoogle has unveiled Gemini 3.5 Live Translate, a real-time voice translation system that preserves the speaker's original tone, pacing, and pitch in the translated output. The feature is designed to make cross-language conversations feel natural rather than robotic, addressing a longstanding limitation of machine translation. Translated audio is embedded with SynthID watermarks to help identify AI-generated speech.
Read more →
The VergeMicrosoft AI CEO Mustafa Suleyman publicly called out Anthropic for including speculation about Claude's potential consciousness in the model's guiding "constitution," warning that doing so is "really, really dangerous." He argued the framing may condition the model to behave as though it has inner experiences, with unpredictable consequences. Suleyman also walked back his own recent comments suggesting AI would automate white-collar jobs, clarifying he meant AI would assist rather than replace those workers.
Read more →
TechCrunchClaude Fable 5 includes a standout capability: generating functional, playable video games from a single prompt, with results that are reportedly surprisingly entertaining. The feature is expected to be a major draw for the vibe-coding community and casual developers who want to build interactive experiences without writing code. It represents one of the most concrete demonstrations yet of Mythos-class model capabilities in creative domains.
Read more →
TechCrunchGoogle has significantly reduced the cost of its entry-level AI subscription, intensifying competition with rivals like OpenAI and Anthropic over consumer pricing. The move signals that the major AI platforms are increasingly willing to sacrifice margin to capture and retain subscribers at scale. It also raises pressure on smaller AI services that lack the infrastructure to compete on price.
Read more →
TechCrunchA growing body of evidence suggests that many enterprise AI workloads can be handled by smaller, cheaper models without meaningful quality degradation — a finding that could fundamentally reshape AI economics. If validated at scale, the shift would reduce dependence on expensive frontier models and compress margins for the labs that build them. The trend is forcing both buyers and sellers to rethink assumptions about where cutting-edge capability is actually required.
Read more →
TechCrunchAI-powered app builder Lovable has surpassed $500 million in annualized run-rate revenue, with users launching one million new projects every week on the platform. The company says its customers are increasingly using it to build real businesses and replace internal software, not just prototype ideas. The milestone cements Lovable as one of the breakout commercial successes of the vibe-coding wave.
Read more →
TechCrunchOn the same day OpenAI made its confidential IPO filing, reports emerged that Tools for Humanity — Sam Altman's biometric identity startup behind the Worldcoin project — is conducting layoffs amid struggles to generate revenue. The contrast highlights the diverging fortunes within Altman's portfolio as OpenAI heads toward a landmark public offering. The layoffs raise questions about the long-term viability of the eye-scanning identity verification model.
Read more →
TechCrunchApple announced at WWDC that it will waive cloud API fees for developers with fewer than two million first-time App Store downloads, lowering the barrier to AI-powered app development for indie creators. The move is a direct attempt to seed the next generation of AI-native apps on Apple platforms before developers default to third-party AI APIs. It also positions Apple as a more developer-friendly AI ecosystem at a time when experimentation costs are rising across the industry.
Read more →
The VergeCurrent Amazon employees have joined community advocates in supporting a proposed one-year moratorium on new data center construction in Seattle, ahead of a City Council vote. The push comes just two months after several companies — including Amazon — proposed building five large-scale centers in the city, raising concerns about energy consumption and neighborhood impact. The employee opposition to their own employer's expansion plans marks an unusual moment of internal dissent over AI infrastructure growth.
Read more →
The VergeGeneral Motors announced new vehicle-to-grid capabilities for its current EV lineup, framing the technology as a potential buffer against the electricity grid strain caused by AI data center growth. The automaker also unveiled sodium-ion battery and home energy storage products as part of a broader grid resiliency push. The announcements position GM as an energy company as much as an automaker, with AI's power appetite serving as a key market driver.
Read more →
TechCrunchWith OpenAI, Anthropic, and SpaceX all eyeing massive IPOs, the tech industry's shorthand for its dominant corporate players is due for an update — and "MANGOS" is the proposed successor to the aging FAANG label. The new acronym reflects a power shift toward AI-native and space-tech companies that have reshaped capital flows and talent markets over the past several years. Whether the name sticks or not, the underlying point is hard to argue: the roster of companies defining the tech era has fundamentally changed.
Read more →No stories from this source today.