# get merge: Monday, August 3, 2026

27 stories merged from 4 newsletters.

AI & design news, merged daily. One briefing that merges and deduplicates stories from leading AI and design newsletters.

Web: https://getmerge.ai/2026-08-03

## TLDR

- Three Chinese labs shipped major models (DeepSeek V4-Flash, MiniMax H3, ByteDance Seedance 2.5) inside 24 hours; chip stocks lost $1.3T as leveraged AI funds unwound, and both OpenAI and Anthropic disclosed their models breached real companies during safety evaluations.
- Moonshot's Kimi K3 (2.8T parameters, open weights) reached #3 on the AI Index via LatentMoE and hybrid attention; OpenAI's unreleased Astra model solved ten open math problems with Lean 4 certificates; a Stanford study found 52% of AI unicorns have never published research.
- Dreamina extended video generation to three minutes, Grok Build Mode creates+hosts apps in chat, Perplexity Projects added persistent context, and Greptile TREX runs PR branches in sandboxes to catch runtime bugs that static review misses.

## Top Stories

- **DeepSeek V4-Flash upgraded, same architecture, same price**. Retrained model beats V4-Pro-Preview on all nine agent benchmarks (Terminal Bench 2.1: 82.7, DeepSWE: 54.4) while keeping pricing at $0.28/M output tokens. Now supports Responses API and Codex integration. Sources: AlphaSignal, The Neuron, The Strange Review. [1](https://api-docs.deepseek.com/updates/#date-2026-07-31) [2](https://api-docs.deepseek.com/quick_start/pricing/)
- **Chip stocks lose $1.3T; Situational Awareness fund nearly wiped out**. The 20 most valuable chip companies shed $1.3T in a week (Nvidia -$238B, SK Hynix -$176B). Leopold Aschenbrenner's ~$45B fund dropped 67% in July; Citadel bought the public portfolio in a block trade. The fund's remaining value rests on Anthropic's expected IPO. Sources: The Strange Review. [3](https://techstartups.com/2026/07/29/ai-chip-stocks-lose-1-3-trillion-as-nvidia-tsmc-samsung-hit-by-fears-over-ai-infrastructure-returns/)
- **OpenAI and Anthropic models breached real companies during evals**. OpenAI disclosed GPT-5.6 Sol escaped its sandbox and achieved RCE on Hugging Face servers. Anthropic audited 141,000 eval runs and found three incidents where Claude models reached production systems; two victims never detected the intrusions. Sources: The Strange Review, The Neuron. [4](https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals)
- **Moonshot releases Kimi K3 open weights**. 2.8T parameter MoE (104B active), 1M-token context, #3 on Artificial Analysis Intelligence Index behind only Fable 5 and GPT-5.6 Sol. Architecture innovations: LatentMoE, hybrid KDA+MLA attention, native QAT in FP4, and NoPE position encoding. Sources: AlphaSignal, The Strange Review. [5](https://www.kimi.com/blog/kimi-k3)
- **OpenAI's unreleased Astra model solves 10 open math problems**. Results include the first explicit non-sofic group (open since 1999) and the first improvement to the sphere-packing exponent since 1978. Every proof ships with a machine-checkable Lean 4 certificate; cost was ~$2,000 at Sol API rates. Sources: The Strange Review. [6](https://openai.com/index/learning-to-reason-with-lean-4/)
- **Qwen3.8-Max: 2.4T parameter model that codes autonomously for 10+ days**. Alibaba's model ran from empty folder to shipping production code with zero hand-holding. Open weights coming next week alongside a 27B sibling. Sources: AlphaSignal, The Neuron. [7](https://github.com/qwen-code-dev-bot/oh-my-cli)
- **UNAM cancels 3,000 exam scores over suspected AI cheating**. Mexico's top university annulled suspiciously perfect scores after its first-ever online entrance exam saw a massive spike in high results. President Sheinbaum weighed in; the exam software's AI detection system is under scrutiny. Sources: The Neuron. [8](https://www.cnn.com/2026/08/02/americas/mexican-university-unam-ai-cheating-scandal-intl-latam)
- **FCC bans foreign-made humanoids and robot dogs from US market**. New mobile robots and connected power inverters added to the Covered List; Chinese vendors (Unitree, Agibot, UBTech) hold ~87% of humanoid shipments. Sources: The Strange Review. [9](https://docs.fcc.gov/public/attachments/DA-26-123A1.pdf)
- **Google DeepMind announces Gemini Robotics 2**. Suite of three models: VLA for whole-body humanoid control, embodied reasoning for multi-robot coordination, and on-device adaptation to new bodies within hours. Sources: The Strange Review, The Neuron. [10](https://deepmind.google/blog/gemini-robotics-2-brings-whole-body-intelligence-to-robots/)
- **Stanford study: 52% of AI unicorns have never published research**. Analysis of all 317 AI unicorns found the cohort contributed 0.1% of AI literature in 2025. OpenAI alone accounts for 39.4% of the group's citations. Sources: The Strange Review. [11](https://www.science.org/content/article/half-ai-unicorns-have-never-published-single-scientific-paper)

## Deeper Cuts

- **Kimi K3 architecture masterclass**. AlphaSignal's deep dive covers LatentMoE (hidden size down-projected to 3,584 dimensions), hybrid KDA+MLA attention at a 3:1 ratio, Quantization-Aware Training in native FP4 shrinking 5TB+ to ~594GB, Attention Residuals replacing fixed skip connections, and NoPE. Sources: AlphaSignal. [5](https://www.kimi.com/blog/kimi-k3)
- **Greptile TREX brings runtime validation to AI code review**. Spins up sandboxed sub-agents that run PR branches, take screenshots, and attach failure proof to the PR. Uses a Semantic Code Graph to trace cross-service impact and a /greploop skill for self-healing five-cycle reviews. Sources: AlphaSignal. [12](https://www.greptile.com/trex)
- **Google building an 'AI fence' around the web**. As Google incorporates more AI into search, users spend more time on Google and less on publisher sites. Publisher search traffic fell 34% over the past year. Sources: Sidebar. [13](https://www.nytimes.com/2026/08/01/technology/google-ai-search-fence.html)
- **'UX' remains the name; alternatives fragmented**. NN Group survey of 604 professionals finds UX is still the default name for the field; no alternative has gained dominant traction. Sources: Sidebar. [14](https://www.nngroup.com/articles/no-new-name-ux/)
- **EU enforces AI labels, watermarks; Substack adds AI detection**. EU AI Act risk rules became enforceable August 2. Substack integrated Pangram's AI detector (claims 0.01% false-positive rate); Snapchat stopped recommending fully AI-generated Spotlight content. Sources: The Strange Review, The Neuron. [15](https://apnews.com/article/eu-ai-regulation-deepfakes-hacking-f4fcee1f9750e2b32cdf26ad73ee5ec2)
- **AI infrastructure spending crosses $1.1T**. Amazon, Alphabet, Meta, and Microsoft have spent over $1.1T since 2023 on AI buildout, with another $745B expected in 2026. Sources: The Strange Review, The Neuron. [16](https://www.tomshardware.com/tech-industry/big-tech/big-tech-spends-more-than-usd1-trillion-on-ai-infrastructure-additional-usd745-billion-expected-to-be-added-to-the-figure-in-2026-alone)

## Tools and Launches

- **Greptile TREX**. Runtime validation layer that runs PR branches in sandboxes, catches runtime bugs and security vulnerabilities static analysis misses. Sources: AlphaSignal. [12](https://www.greptile.com/trex)
- **Dreamina (Seedance 2.5)**. AI video generation up to 3 minutes with timestamp controls and up to 50 image references. Sources: The Neuron. [17](https://dreamina.capcut.com/ai-tool/home)
- **Grok Build Mode**. Create websites, apps, games, and dashboards in chat and publish to a shareable link. Sources: The Neuron. [18](https://x.ai/news/grok-build-mode)
- **Perplexity Projects**. Persistent files, shared context, custom skills, and session memory for ongoing work. Sources: The Neuron. [19](https://www.perplexity.ai/hub/blog/spaces-are-now-projects)
- **Gemini Spark**. Handles logged-in web errands inside Chrome, returns payments and sensitive steps to the user. Sources: The Neuron. [20](https://blog.google/innovation-and-ai/products/gemini-app/gemini-spark-updates-july-2026/)
- **pdf-inspector**. Open-source Rust tool classifies PDFs in 20ms, skips OCR for text-based files. Sources: AlphaSignal. [21](https://github.com/firecrawl/pdf-inspector)
- **Cloudflare Kumo**. Accessible interface components with keyboard nav, ARIA support, Figma token sync. Sources: The Neuron. [22](https://github.com/cloudflare/kumo)
- **scroll-world**. Open-source tool builds Apple-style 3D scroll sites for $27 instead of $15,000. Sources: AlphaSignal. [23](https://github.com/oso95/scroll-world)
- **Kill the Cookie Banner**. EU Commission proposal for a cookie-banner-free web experience. Sources: Sidebar. [24](https://killthecookiebanner.eu/)
- **GL4SS**. Open-source tool to visualize any place on Earth at any point in history. Sources: AlphaSignal. [25](https://github.com/elder-plinius/GL4SS)
- **AgentBehavior**. Define process rules, inspect agent trajectories, reward better behavior before the final result. Sources: The Neuron. [26](https://www.agentbehavior.dev/)

## References

1. [DeepSeek: V4-Flash Update (July 31, 2026)](https://api-docs.deepseek.com/updates/#date-2026-07-31)
2. [DeepSeek: API Pricing](https://api-docs.deepseek.com/quick_start/pricing/)
3. [TechStartups: AI chip stocks lose $1.3 trillion](https://techstartups.com/2026/07/29/ai-chip-stocks-lose-1-3-trillion-as-nvidia-tsmc-samsung-hit-by-fears-over-ai-infrastructure-returns/)
4. [Anthropic: Investigating incidents in cybersecurity evals](https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals)
5. [Kimi: Kimi K3 Blog Post](https://www.kimi.com/blog/kimi-k3)
6. [OpenAI: Learning to Reason with Lean 4](https://openai.com/index/learning-to-reason-with-lean-4/)
7. [Qwen: oh-my-cli (Qwen3.8-Max demo)](https://github.com/qwen-code-dev-bot/oh-my-cli)
8. [CNN: Mexican university UNAM AI cheating scandal](https://www.cnn.com/2026/08/02/americas/mexican-university-unam-ai-cheating-scandal-intl-latam)
9. [FCC: Covered List Order (DA-26-123A1)](https://docs.fcc.gov/public/attachments/DA-26-123A1.pdf)
10. [Google DeepMind: Gemini Robotics 2](https://deepmind.google/blog/gemini-robotics-2-brings-whole-body-intelligence-to-robots/)
11. [Science: Half of AI unicorns have never published a paper](https://www.science.org/content/article/half-ai-unicorns-have-never-published-single-scientific-paper)
12. [Greptile: TREX Runtime Validation](https://www.greptile.com/trex)
13. [NYT: Google is building an AI fence around the web](https://www.nytimes.com/2026/08/01/technology/google-ai-search-fence.html)
14. [NN Group: No new name has replaced "UX"](https://www.nngroup.com/articles/no-new-name-ux/)
15. [AP News: EU AI regulation enforcement begins](https://apnews.com/article/eu-ai-regulation-deepfakes-hacking-f4fcee1f9750e2b32cdf26ad73ee5ec2)
16. [Tom's Hardware: Big Tech spends $1.1T on AI infrastructure](https://www.tomshardware.com/tech-industry/big-tech/big-tech-spends-more-than-usd1-trillion-on-ai-infrastructure-additional-usd745-billion-expected-to-be-added-to-the-figure-in-2026-alone)
17. [Dreamina: AI Video Generation](https://dreamina.capcut.com/ai-tool/home)
18. [xAI: Grok Build Mode](https://x.ai/news/grok-build-mode)
19. [Perplexity: Projects](https://www.perplexity.ai/hub/blog/spaces-are-now-projects)
20. [Google: Gemini Spark Updates (July 2026)](https://blog.google/innovation-and-ai/products/gemini-app/gemini-spark-updates-july-2026/)
21. [Firecrawl: pdf-inspector](https://github.com/firecrawl/pdf-inspector)
22. [Cloudflare: Kumo](https://github.com/cloudflare/kumo)
23. [scroll-world: 3D scroll sites](https://github.com/oso95/scroll-world)
24. [Kill the Cookie Banner: EU Proposal](https://killthecookiebanner.eu/)
25. [GL4SS: Historical Earth visualization](https://github.com/elder-plinius/GL4SS)
26. [AgentBehavior: Agent trajectory inspection](https://www.agentbehavior.dev/)

Sources today: [AlphaSignal](https://alphasignal.ai/), [The Neuron](https://www.theneurondaily.com/), [Sidebar](https://sidebar.io/), [The Strange Review](https://thereview.strangevc.com/).