AI News Brief | 2026-06-16
Anthropic's Claude Mythos 5 and Fable 5 face global suspension amid US export controls, while DeepSeek secures a massive $7B funding round and Midjourney V8.1 becomes the default model.
## AI Industry Overview
- **Anthropic vs. US Government**: The Trump administration issued an export control directive suspending global access to Anthropic’s Claude Fable 5 and Mythos 5 over national security and alleged jailbreak concerns. Anthropic is actively engaging the White House to lift the restrictions, while the Pentagon shifts its daily AI workflows away from the lab. ([The Verge](https://www.theverge.com/ai-artificial-intelligence/950412/anthropic-trump-adminstration-claude-mythos-fable-5-export-controls))
- **DeepSeek's Record Funding**: Chinese AI lab DeepSeek has completed a massive financing round exceeding $7 billion, valuing the company at over $50 billion and intensifying the global AI race. ([The Information](https://www.ainvest.com/news/information-deepseek-completed-record-financing-exceeding-7-0-billion-valuing-company-50-0-billion-2606/))
- **OpenAI Multistate Probe**: Ahead of its looming IPO, OpenAI is facing a multistate probe by attorneys general investigating potential user harm and scrutinizing ChatGPT's safety claims. ([TechXplore](https://techxplore.com/news/2026-06-openai-multistate-probe-user-ipo.html))
- **Salesforce Acquires Fin**: Salesforce announced a $3.6 billion acquisition of AI customer service platform Fin to bolster its enterprise Agentforce platform. ([TechCrunch](https://techcrunch.com/2026/06/15/salesforce-acquires-ai-customer-service-platform-fin-for-3-6b))
- **Nvidia's AI Debt Boom**: Nvidia plans to raise at least $20 billion through its first bond sale since 2021, joining the growing trend of debt financing in the AI sector. ([The Decoder](https://the-decoder.com/nvidia-joins-ai-debt-boom-with-20-billion-bond-sale))
- **KAIST Cooling Breakthrough**: Researchers at KAIST unveiled a liquid cooling technology that is 10x more efficient than previous methods, addressing the massive power consumption of AI data centers. ([ScienMag](https://scienmag.com/kaist-shatters-ai-cooling-barrier-with-liquid-technology-10x-more-efficient-than-previous-best/))
## Tool Updates
### Claude
Anthropic has been forced to globally suspend access to its newest models, Claude Fable 5 and Mythos 5, following a US government export control directive citing national security concerns. The company is currently in discussions with the Trump administration to reverse the ban and restore access for foreign nationals. ([Anthropic](https://www.anthropic.com/news/fable-mythos-access?rd=1))
### ChatGPT
OpenAI is under intense scrutiny as a multistate probe investigates potential user harm and safety claims ahead of the company's IPO. Additionally, OpenAI announced a deadline calendar for the retirement of older models, including GPT-5.2, GPT-4.5, and o3. ([TechScurrent](https://techscurrent.com/2026/06/openai-chatgpt-state-attorneys-general-probe/))
### DeepSeek
DeepSeek has closed a record-breaking $7.4 billion funding round, pushing its valuation past $50 billion. This massive capital injection highlights the startup's continued disruption of the AI market, accompanied by aggressive API price collapses. ([AI in China](https://www.ainchina.com/blog/deepseek-7-billion-fundraising-tencent-catl-api-price-war-2026/))
### Devin & Windsurf
Cognition has officially rebranded Windsurf to Devin Desktop, transforming the IDE into an agent command center with the new Agent Client Protocol (ACP). This strategic pivot follows Cognition's massive $1 billion fundraising round at a $26 billion valuation. ([The Agent Report](https://the-agent-report.com/2026/06/cognition-devin-desktop-agent-orchestration/))
### Qwen
Alibaba's Qwen team has expanded beyond chatbots by launching a suite of robotics models, including Qwen-RobotManip, Qwen-RobotWorld, and Qwen-RobotNav. These models unify visual-language-action frameworks for advanced robotic manipulation, navigation, and autonomous driving. ([Qwen Blog](https://qwen.ai/blog?id=qwen-robotmanip))
### Midjourney
Midjourney V8.1 is now the default model, offering smarter prompt adherence, better text rendering, and significantly faster generation times. With HD mode enabled, V8.1 renders at 4x the resolution of V7 images while maintaining rapid output speeds. ([Midjourney](https://updates.midjourney.com/v8-1-is-now-the-default-model/))
### Cursor
Cursor 3.7 has shipped with major upgrades to Design Mode, allowing developers to edit UI elements using voice commands and multi-select pointing. This update streamlines frontend workflows by letting users queue the next edit while the agent is still working. ([Vibe Coder Blog](https://blog.vibecoder.me/cursor-design-mode-visual-ui-agent-edits))
### Claude Code
Anthropic introduced five-level nested sub-agents in Claude Code, allowing agents to spawn their own sub-agents for complex orchestration. The update also brings dynamic workflows, enabling the tool to write and manage its own multi-agent harness on the fly. ([Claude Blog](https://claude.com/blog/a-harness-for-every-task-dynamic-workflows-in-claude-code))
### GitHub Copilot
GitHub Copilot CLI has been updated to be more selective about task delegation, preventing unnecessary helper agent spin-ups for simple changes. A new `/settings` command also provides a unified, schema-driven interface for configuring the CLI environment. ([GitHub Blog](https://github.blog/ai-and-ml/how-we-made-github-copilot-cli-more-selective-about-delegation/))
### Gemini
Google confirmed that Gemini 3.5 Pro will launch in June 2026, featuring a massive 2-million-token context window to target frontier-model leadership. Meanwhile, Gemini 3.5 Live Translate is now available, offering near real-time speech-to-speech translation in over 70 languages. ([Google Blog](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-live-3-5-translate/))
### Kimi
Moonshot AI has launched and open-sourced Kimi K2.7 Code, a high-speed coding model that outputs up to 6x faster than its predecessor. The model is highly cost-effective and demonstrates significant improvements in long-context programming and instruction following. ([Kimi](https://mp.weixin.qq.com/s/p87ebkY1xqKtkGZ2N3DGSw))
### ChatGLM
Z.ai (Zhipu AI) released GLM-5.2, featuring a fully usable 1-million-token context window and two thinking-effort levels optimized for agentic coding. The model is currently available across all tiers, with MIT-licensed open weights expected to drop next week. ([MarkTechPost](https://www.marktechpost.com/2026/06/14/z-ai-launches-glm-5-2-with-a-usable-1m-token-context-two-thinking-effort-levels-and-no-benchmarks-at-launch/))
### Kling
Kling AI has launched its 3.0 model, introducing industry-first native 4K video generation that meets rigorous cinematic production standards. This update empowers creators to direct high-fidelity, multi-shot videos directly from text prompts. ([Financial Content](https://www.financialcontent.com/article/getnews-2026-6-12-kling-ai-launches-30-model-ushering-in-an-era-where-everyone-can-be-a-director))
### ElevenLabs
ElevenLabs introduced Avatars in ElevenCreative, enabling users to generate talking-head videos with integrated voice and lip-sync in a single workflow. The company also launched Dubbing v2, a revolutionary model designed to streamline enterprise video localization. ([ElevenLabs](https://elevenlabs.io/blog/introducing-avatars))
### Perplexity
Perplexity AI unveiled a hybrid local-cloud inference orchestrator that dynamically routes tasks between PCs and cloud servers to drastically cut compute costs. Additionally, Deep Research has been upgraded to route subtasks across 20+ frontier models for comprehensive reports and dashboards. ([VentureBeat](https://venturebeat.com/technology/perplexity-ai-unveils-hybrid-local-cloud-inference-system-at-computex-2026))
### Adobe Firefly
Adobe's June 2026 Creative Cloud updates bring enhanced AI user controls to Lightroom and Photoshop, streamlining the culling and compositing workflows. Notably, Lightroom can now leverage generative AI to turn static photos into dynamic videos. ([PetaPixel](https://petapixel.com/2026/06/15/adobe-adds-more-user-control-to-ai-features-inside-lightroom-and-photoshop/))
### Pika
Pika has launched PikaStream, a beta feature that enables real-time video chat with AI agents at 24fps with just 1.5-second latency. This builds upon their Pika Agents initiative, shifting generative video from one-off prompts to persistent, conversational partnerships. ([AI News Lab](https://www.ainewslab.org/en/article/pika-pikastream-realtime-video-agents))
### Flux (Black Forest Labs)
Legendary film director Martin Scorsese has joined Black Forest Labs, the startup behind the FLUX image model, as an adviser and partner. Scorsese will utilize the AI platform to assist in storyboarding and visualizing cinematic concepts. ([Rolling Stone](https://www.rollingstone.com/tv-movies/tv-movie-news/martin-scorsese-ai-company-black-forest-labs-storyboards-1235571356/))
### Notion AI
Notion experienced a brief service disruption over the weekend that took its Anthropic Claude integration offline for several hours. The company has since restored full access to Opus 4.7 and 4.8 models for its AI assistant features. ([TechCrunch](https://techcrunch.com/2026/06/07/notion-restores-access-to-anthropic-after-service-disruption/))
### v0
Vercel's v0 is evolving from a UI generator into a full agentic coding platform capable of resolving PR merge conflicts and writing SQL. A new context-aware command palette (Cmd+K) allows developers to jump seamlessly between chats, projects, and actions. ([v0 Changelog](https://v0.dev/changelog))
### Aider
A critical SQL injection vulnerability (CVE-2026-10176) was disclosed in Aider-AI's code generation workflow, exposing servers running version 0.86.3 or earlier to remote attacks. The latest releases have patched this flaw while adding support for GPT-5 and Grok-4 models. ([Webmastered](https://www.webmastered.com/blog/sql-injection-code-generation-tool-vulnerability/))
### Cline
LG CNS and Cline have partnered to launch "Cline Spec Driven for Enterprise," an agentic AI solution designed to orchestrate the full software development lifecycle. This initiative brings intelligent automation from initial analysis and design straight through to coding and QA. ([PR Newswire](https://www.prnewswire.com/news-releases/going-beyond-vibe-coding-agentic-ai-takes-on-large-scale-system-development-302791263.html))
### Cody
Sourcegraph has shifted Cody to an enterprise-only model, moving away from its previous free-tier-friendly approach for indie developers. The AI code assistant now focuses strictly on helping large organizations achieve codebase consistency and quality at scale. ([DEV Community](https://dev.to/jovan_chan_9500711396d4e6/sourcegraph-cody-review-2026-enterprise-only-now-and-what-that-means-for-indie-devs-1d0))
### AKOOL
AKOOL has launched Agentic Canvas, a visual execution layer where intelligent agents plan, coordinate, and execute complex business and media workflows. This update aims to redefine professional video content creation by integrating multi-agent orchestration. ([PR Newswire](https://www.prnewswire.com/news-releases/akool-launches-agentic-canvas-to-redefine-how-work-gets-done-with-ai-302795966.html))
### BACH
Singapore-based Video Rebirth has launched BACH 1.0, an AI video engine capable of directing 30-second multi-shot films from a single prompt. The tool boasts true character consistency and cinematic fidelity, ranking highly on artificial analysis benchmarks upon debut. ([BACH Video AI](https://bachvideoai.com/))
### Doubao
ByteDance’s AI assistant Doubao is preparing to launch paid subscription services in late June, signaling the end of its free-first growth phase. The monetization push will be closely integrated with Douyin's e-commerce ecosystem to drive AI-powered shopping experiences. ([TechNode](https://technode.com/2026/06/02/bytedances-doubao-to-launch-paid-plans-in-late-june-link-with-douyin-e-commerce-push/))
### Sora
OpenAI has officially shut down its Sora text-to-video app and API, citing unsustainable economics and poor user retention rates. The shutdown highlights the immense compute costs and current financial hurdles of scaling AI video generation for consumer markets. ([ngram](https://www.ngram.com/blog/openai-sora-shutdown-ai-video-economics))