Artificial Curiosity Labs
Go back

AI News Digest

AI News Digest — 2026-08-13

Top Stories

Lovable confirms new $13.3B valuation, raises another $400M (TechCrunch)

Lovable’s new $13.3 billion valuation and $400 million raise come as the company reaches $500 million in annual recurring revenue, placing it among the highest-valued AI startups and underscoring that investors are still aggressively backing AI application companies at scale.

AI coding startup Cognition reportedly already in talks to raise at $40B valuation (TechCrunch)

Cognition’s reported move to raise at a $40 billion valuation, up from $26 billion, signals intense investor demand for AI coding agents and may set a new benchmark for the category.

DeepSeek V4 Pro Goes Live, Closes In on Frontier Tier (Pandaily)

DeepSeek V4 Pro’s launch pushes open-weight models closer to the frontier tier by pairing a 1M-token context window with explicit agent tooling, a combination that raises competitive pressure on established labs and lowers the barrier for developers building long-context, tool-using AI systems.

SpaceXAI debuts Grok 4.6, overtaking Kimi K3 and matching GPT-5.6 Sol for world’s third best (VentureBeat)

SpaceXAI’s Grok 4.6 has vaulted into a tie for third place with GPT-5.6 Sol on the Artificial Analysis leaderboard while specifically surpassing Kimi K3, a notable shift given Grok’s lower positioning just a quarter ago; its competitive pricing could intensify pressure on both proprietary and open-source labs as agentic and coding use cases become primary battlegrounds.

Made by Google 2026: all the Pixel news and announcements (The Verge)

Google’s Pixel 11 lineup showcases a shift where foundation models like Gemini AI become a central, hardware-level feature, with integration across devices and a visible LED ring indicator underlining that on-device intelligence is now a core, persistent part of the user experience rather than a background service.

Funding & Capital

What moved

Lovable confirms new $13.3B valuation, raises another $400M (TechCrunch)

Lovable’s confirmation of a $13.3 billion valuation and a $400 million raise, with annual recurring revenue reaching $500 million, places it among the highest-valued AI startups and signals that investors remain eager to back AI application companies with strong revenue traction.

Huya’s Game Publishing Bet Starts to Pay Off as Q2 Revenue Rises 11% (pandaily.com)

Huya’s second-quarter results show its game publishing bet is paying off, with gaming revenue surging 54% and total revenue rising 11% year-over-year, while the board’s move to double its share buyback program signals strong internal confidence in the platform’s accelerating momentum.

AI coding startup Cognition reportedly already in talks to raise at $40B valuation (TechCrunch)

Cognition is reportedly in discussions to raise new funding at a $40 billion valuation, a sharp climb from $26 billion. The jump highlights accelerating investor competition for AI coding tools and could reset valuation expectations across the category.

Also tracking

Foundation Models

What moved

Tencent Plans Larger Hy4 Model After Hy3 Usage Jumps 68-Fold (TechNode)

Tencent’s surging internal adoption of its Hunyuan Hy3 model, with usage leaping 68-fold, directly fuels the accelerated push for a larger, more capable Hy4, underscoring that the firm is competing on raw model scale and real-world deployment utility rather than merely hyping research benchmarks.

WeChat AI Team Details WeLM Models Scaling to 617B Parameters (TechNode)

The WeChat AI team demonstrated that their novel Hidden Decoding training method allows WeLM models to scale up to 617 billion parameters without needing to expand the underlying Transformer backbone, representing a noteworthy departure from the usual parameter-versus-infrastructure tradeoff.

DeepSeek V4 Pro Goes Live, Closes In on Frontier Tier (Pandaily)

DeepSeek V4 Pro’s launch puts a frontier-tier open-weight model into the market with 1M-token context and explicit agent tooling, directly intensifying competition at the top of the open-weight LLM landscape.

Also tracking

Agents & Tooling

What moved

Alibaba Cloud Launches Qwen AI Arena for Real-World Agent Testing (TechNode)

Alibaba Cloud’s introduction of the Qwen AI Arena provides a dedicated environment for testing AI agents against real business scenarios, moving evaluation beyond static language model leaderboards to address the practical reliability and task-completion metrics that enterprises require before deployment.

‘Specialists aren’t required’ anymore: How to stay valuable in an AI agent workplace today (ZDNet)

AI agents are reshaping workforce expectations, shifting demand away from narrow specialists and toward versatile generalists who can work across the tech stack, making cross-disciplinary adaptability central to staying valuable.

Scaling AI agents with trustworthy data (MIT Technology Review)

The return on investment from agentic AI is fundamentally dependent on trustworthy data, and this dependency is rapidly becoming a critical bottleneck as enterprises accelerate their adoption of AI agents.

Also tracking

Applications

What moved

WPS Quietly Launches Lingxi Pro, an AI-Native Office Agent That Turns Skills into Editable Documents (pandaily.com)

WPS quietly launched Lingxi Pro, an AI-native office agent that inverts the traditional document-first paradigm by turning skills directly into editable documents, a shift now exposed to its 678 million monthly active users.

Tencent’s Q2 2026 Earnings Call: Three-Layer AI Strategy — Intelligence, Applications, Infrastructure — Shows Traction (pandaily.com)

Tencent’s Q2 2026 earnings call detailed a three-layer AI strategy spanning intelligence, applications, and infrastructure, pointing to Hunyuan Hy3, WorkBuddy, and CodeBuddy as concrete proof points, which makes the update a useful signal of how a major platform operator is turning its AI roadmap into shipped products.

Mesh, Automattic’s CRM for everyone, comes to Android (TechCrunch)

Automattic’s expansion of its AI‑powered CRM Mesh to Android signals a strategic push to embed adaptive relationship‑management tools directly into users’ most personal device, normalizing AI as a persistent background assistant for everyday interactions rather than a specialized business function.

Also tracking

Image & Video

What moved

Some Claude users are mad that Anthropic’s new watermarks will catch them using it at their jobs, classes (TechCrunch)

Anthropic’s rollout of output watermarking for Claude has provoked user backlash over fears it will expose AI use in workplace and academic settings, directly elevating the stakes for AI-content detection and challenging existing norms of academic integrity and enterprise content policies.

Guitar company D’Addario admits that AI music was used in a promotional video (The Verge)

D’Addario’s acknowledgment that AI-generated music was used in a promotional video without upfront disclosure highlights the mounting creative-industry tension over authenticity and transparency, as a major music-adjacent brand quietly blending synthetic audio into its marketing raises questions about audience trust and the standards brands should meet when mixing AI content with traditional craft.

4 New Camera Tricks on Google’s Latest Pixel 11 Smartphones (WIRED)

Google’s Pixel 11 leverages on-device AI with features like Magic Capture and Instant Night Sight, which process complex imaging tasks locally without cloud dependency, reinforcing the trend toward powerful edge AI directly in consumer handsets to improve speed and privacy.

Also tracking

Policy & Safety / Other

What moved

Twitch streamers can now opt out from training Amazon’s AI (The Verge)

Twitch’s opt-out toggle formalizes a tension between platform control and creator consent, as parent company Amazon signals its intent to feed streamer content into AI models unless individuals actively block it, leaving “opt-out by default” as the practical norm that shifts the burden onto creators and raises pressure for clearer data-rights standards across user-generated content platforms.

As AI safety concerns mount, three pioneers make the case for staying open (TechCrunch)

The open-versus-closed debate over AI models took center stage as pioneers Geoffrey Hinton, Fei-Fei Li, and Andrew Ng clashed at a major industry conference, surfacing the fundamental policy tension between accelerating open-source access and imposing safety-driven regulation.

The web’s newest weapon against AI scrapers is a font (Ars Technica)

ShieldFont offers a grassroots data-poisoning defense by turning ordinary webpage text into nonsense for AI scrapers, a novel countermeasure as tensions over AI training scraping escalate.

Also tracking

Voice & Speech

What moved

Why Sandbar thinks its voice-enabled ring can avoid the AI hardware graveyard (TechCrunch)

Sandbar’s voice-enabled ring pushes into the crowded AI wearable space with a form factor distinct from the pins and pendants that have struggled to gain traction, testing whether a voice-notetaking device worn on the finger can find durable demand where earlier novel hardware has not.

Robotics & Physical AI

What moved

Gemini Robotics 2 brings whole body intelligence to robots (Google DeepMind Blog)

Google DeepMind’s Gemini Robotics 2 introduces whole-body intelligence, moving robots beyond arm-only manipulation to coordinated full-body actions, marking a significant robotics milestone first reported in July 2026.

Optical Tech Would Update a Robot’s AI on the Fly (IEEE Spectrum)

A new optical communication method enables real-time, over-the-air updates to a robot’s AI model, directly transferring weight data to its memory without pausing operations, according to IEEE Spectrum. This matters because it solves a key deployment bottleneck: rather than pulling robots offline for manual software updates—an impractical approach for large, dispersed fleets—the technology allows continuous adaptation to new tasks or environments, drastically reducing maintenance overhead and making persistent, life-long learning feasible for physical AI systems in the field.

Infra & Compute

What moved

AI chipmaker Cerebras Systems’ stock plunges, despite posting solid earnings and guidance (SiliconANGLE)

Cerebras Systems shares fell sharply even after the AI chipmaker reported earnings above expectations and lifted its outlook, underscoring a disconnect between solid AI compute demand and investor caution toward AI hardware names.

Nebius shares jump 34% on continued AI infrastructure demand (SiliconANGLE)

Nebius shares jumped 34% after the AI-focused cloud platform beat across the board, underscoring that demand for specialized AI infrastructure remains red-hot.

AI infrastructure spending boosts Cisco’s earnings and revenue, but its stock declines after-hours (SiliconANGLE)

Cisco’s earnings and revenue got a boost from AI infrastructure spending, but its stock still fell after-hours as margin concerns weighed on the reaction — a sign that while the AI buildout is lifting networking giants and strong guidance suggests the trend has legs, investors are demanding profitability along with growth.

Also tracking