Abu Dhabi Launches World’s First AI Court Platform
ALSO: Moonshot’s Kimi K3 open weights, Patreon cuts 20%
Welcome to Guru99 AI Report!
Top Story: Hi there — this week a courtroom hands AI its case files, a 2.8-trillion-parameter model rattles Washington, and Patreon insists its layoffs weren’t AI’s fault. The gap between what’s said and what’s shipping is widening. Let’s dig in.
⚖️ Abu Dhabi Hands Its Courts an AI Assistant
Brief Buzz:
Abu Dhabi’s judiciary is rolling out what officials bill as the world’s first fully integrated AI court platform — a system that reads case files, surfaces precedents, drafts legal documents, and even offers legal analysis, with human judges keeping the final word.
- The selling point: it bundles legal research, drafting, and analysis into one platform, not separate tools.
- It’s run by the Abu Dhabi Judicial Department — covering Abu Dhabi’s courts, not every court in the UAE.
- Every AI output is human-reviewed and verified; the software can’t issue a ruling on its own.
- Phase one begins in September 2026, with a phased rollout over the following 18 months.
💡 Why Should You Care?
Faster courts could mean quicker verdicts and lower costs. The flip side: AI shaping legal analysis raises real questions about bias, transparency, and accountability.
🌕 Kimi K3’s Weights Are Out — And Washington’s Angry
Brief Buzz:
Moonshot AI just published the full weights for Kimi K3, a 2.8-trillion-parameter model that benchmarks near the closed frontier. It’s the largest open-weight release ever shipped — and it arrived days after the White House accused Moonshot of stealing Anthropic’s tech.
- 2.8 trillion parameters total, but only 16 of 896 experts fire per token — about 50 billion doing real work.
- Moonshot claims a 2.5× scaling-efficiency gain over Kimi K2, crediting new attention and mixture-of-experts architecture.
- It sits just behind Claude Fable 5 and GPT-5.6 Sol on Artificial Analysis’s intelligence index, and topped Arena’s blind frontend-coding vote.
- The White House alleges Moonshot distilled Anthropic’s Fable 5; Treasury floated sanctions. Moonshot denies it; analysts question the 15-day window.
- Xi Jinping backed open-source AI in his first WAIC keynote — one day after K3’s launch.
💡 Why Should You Care?
Open weights mean anyone can self-host frontier AI — but K3 needs roughly 1.4TB of memory, so “free” still favors whoever owns the GPUs.
✂️ Patreon Cuts 20% of Staff, Insists AI Isn’t Why
Brief Buzz:
Patreon just cut 93 jobs — 20% of its workforce — while calling its core business healthy and strong. CEO Jack Conte went out of his way to say AI isn’t replacing anyone, then admitted AI reshaped how the company works anyway.
- 93 employees gone Thursday — a fifth of staff — in Patreon’s deepest cut since 2022, when it shed 17%.
- Conte’s note to creators says AI isn’t replacing staff, yet AI “fundamentally transformed” how Patreon operates.
- Severance runs 16 weeks base pay, a week per year served, year-end healthcare, and a $1,500 laptop stipend.
- Two weeks earlier, Patreon began blocking AI training crawlers with Cloudflare to protect creators’ work.
- Uber cut 10% of support staff a day earlier, naming AI as the reason outright.
💡 Why Should You Care?
Even AI-skeptical companies are shrinking around AI. For workers, the distinction between “replaced by AI” and “reorganized because of AI” may matter less than the outcome.
🧠 Directing AI Beats Knowing More, KPMG Study Finds
Brief Buzz:
Half of junior professionals outperformed an AI agent working alone — and the rest looked identical on paper. A KPMG and UT Austin field study of 523 early-career staff found the difference was how they directed the AI.
- 50.1% were AI Amplifiers — they framed problems, applied domain knowledge, and refined output across rounds, beating the AI-only baseline.
- 25.8% were AI Delegators, taking AI’s output with little scrutiny; they matched AI working alone.
- 24.1% were AI Apprentices — strong on foundational skills, but their critiques misfired and they landed below the baseline.
- Apprentices out-scored Delegators on every foundational skill, so critical thinking and AI literacy alone didn’t predict results.
- Caveat: KPMG co-authored the research and ran it on its own US employees; findings published in Harvard Business Review.
💡 Why Should You Care?
If your employer is pushing AI, the takeaway is that access isn’t enough — rubber-stamping AI output buys nothing. Judgment is what pays, and KPMG argues that gap is coachable.
😻 FLUX 3 Makes Videos — and Powers Factory Robots
Brief Buzz:
Black Forest Labs’ FLUX 3 is not just another prettier-picture machine. The multimodal model generates video with synchronized audio and supplies the foundation for FLUX-mimic, a robotics system being tested on real Audi production tasks.
- 20-second clips can be generated from text, images, existing video, audio, or keyframes, with native sound.
- Multilingual dialogue, animated typography, multiple aspect ratios, and chained clips support more controlled, longer sequences.
- In BFL’s preliminary tests, FLUX 3 was preferred over Runway Gen-4.5 77% of the time and Luma Ray 3.2 93%.
- FLUX-mimic uses the same backbone to handle parts, seals, cables, and assembly work in Audi production environments.
💡 Why Should You Care?
The bigger bet is that one model can power creative tools, simulations, and adaptable robots. But access is limited, and the strongest performance claims remain company-reported.
🌍 Zuckerberg Wants Superintelligence for Everyone, Not a Few Labs
Brief Buzz:
Meta CEO Mark Zuckerberg used a Wall Street Journal op-ed to make his case for superintelligence — AI beyond human capacity — arguing it should be widely distributed, not locked inside a few labs. Expect it within a few years, he writes.
- The defining question, he says, isn’t whether superintelligence arrives but who gets access to it.
- He rejects the claim that safety demands “extreme concentration of power” in a few institutions, calling that the real danger.
- Widely shared AI means more jobs, not fewer, he predicts, with cheaper company-building tilting toward entrepreneurship.
- His three principles: individual empowerment, invention over automation, and balance of power as the foundation of safety.
- The op-ed lands as 1,000+ staff at rival labs urge Washington to slow frontier AI.
💡 Why Should You Care?
His “access for everyone” case doubles as strategy — Meta’s newest models have shifted toward proprietary access, so watch what actually ships, not the rhetoric.
Hey! I’m Krishna Rungta
Founder of Guru99.com, Editor-in-chief & Technology Expert
Was this email forwarded to you? Sign up for free here.
