Good morning. It's Thursday, August 6, and we're covering NVIDIA's push for AI security standards, China's open-model expansion across Africa, and Google's biggest AI leadership reshuffle yet.
Plus: poll on Google's AI leadership shake-up, a guest essay on Jack Dorsey's Buzz, and yesterday's Real vs. AI verdict.
Demis Hassabis steps away from daily management just as Google says AGI is within reach, shifting his focus to a long-term strategy. Google appoints Hassabis as Alphabet’s Chief Scientist and names Koray Kavukcuoglu SVP of Google DeepMind to lead Gemini development and frontier AI research. Gemini surpasses 950 million monthly users, while Gemma exceeds 900 million downloads and Jeff Dean departs after 27 years.
Anthropic joins the race to build custom AI chips instead of relying solely on external hardware, despite development costs that can approach $500 million. The company is assembling an in-house silicon team to co-design chips and Claude models for faster, more efficient inference at scale. The effort complements partnerships with AWS, Google, NVIDIA, and AMD, while Samsung is reportedly under consideration as a manufacturing partner.
Hark claims its new browser agent outperforms frontier AI models on web tasks while costing less to operate. The company unveils Handoff, an AI system designed to complete end-to-end online actions such as booking flights, shopping, ordering food, and conducting research. Hark says Handoff leads three browser-use benchmarks, enters research preview today, and is scheduled for broader access later this summer.
Meta launched Muse Code this week, a terminal coding agent powered by its Muse Spark model that spawns parallel sub-agents to handle complex tasks across large repositories, with Zuckerberg claiming an internal test in which the tool shipped six game features simultaneously. Meta AI chief Alexandr Wang says that Muse Code is meant to be an "incredibly good option, especially from a cost perspective."
POWERED BY TRENDOS.COM
Is Your Brand Visible in AI Search Results?
Most brands have no idea whether AI search engines recommend them or a competitor. Trendos.com tracks your brand across ChatGPT, Perplexity, Gemini, Google AI Overviews, and Claude with real conversational prompts, not keyword proxies. Free to start.
ZEITGEIST POLL
What Does Google’s AI Leadership Shuffle Really Mean?
Google just made major changes to its AI leadership: Demis Hassabis is stepping away from day-to-day management, while Jeff Dean and several other senior researchers are leaving to start a new company.
How should we read Google’s AI shake-up? |
VIDEO
Essential ChatGPT Tips and Tricks
Learn hidden ChatGPT power features—from browser automation and voice agents to scheduled tasks, skills, plugins, and model selection.
FORWARD FUTURE ORIGINAL
Will Jack’s Buzz Attack Sting Slack?
Jack Dorsey tossed a jar of stinging insects into the agentic conversation with Buzz, an open-source workspace where people, code, and AI agents share the same channels. Millions noticed, and many immediately took issue with the data policies around its hosted version. Fair enough.
This is the agentic era: everybody arrives carrying a pitchfork, a threat model, and several strongly held opinions about data sovereignty and risk. → Read the full article here.
SECURITY
NVIDIA-Led Open Secure AI Alliance Launches Security Working Group
Just one week after launching, the NVIDIA-backed Open Secure AI Alliance (OSAA) has unveiled its first initiative: the Shared AI Findings Exchange (SAFE), a working group focused on improving AI cybersecurity practices. Developed during the Black Hat conference in Las Vegas, the initial proposals outline voluntary processes for confidentially reporting AI security incidents, notifying affected parties, and conducting blame-free post-incident reviews.
The alliance has grown to more than 120 members—including Adobe, BlackRock, Cisco, Intel, Microsoft, and Visa—and is also cataloging open source tools from companies such as NVIDIA, Amazon, and Red Hat to help secure AI systems. Notable AI companies including OpenAI, Google, and Anthropic have not joined the alliance, though OpenAI and Google previously signed the open letter that spurred its creation. → Read the full article here.
GEOPOLITICS
China’s Open AI Models Gain Ground Across Africa
Chinese AI models are rapidly gaining adoption across Africa as developers prioritize lower costs, customization, and open access over proprietary U.S. alternatives. Across Uganda, Kenya, Nigeria, and Ghana, startups are using open-source models from companies like Alibaba and DeepSeek to build local-language assistants, legal tools, educational platforms, and business software at a fraction of the cost of commercial U.S. models.
The shift reflects a broader competition for global AI influence, with China pairing its technology offerings with developer support and existing infrastructure investments, while many U.S. companies remain focused on premium, closed models. Even so, developers continue to use American AI for high-performance tasks, highlighting that the global AI race is increasingly defined by tradeoffs between cost, performance, and geopolitical risk rather than a single dominant provider. → Read the full article here. (Paywall)
NEWS
What Else is Happening
Google Eyes $1.5B Mechanize Deal: Google is in talks to license Mechanize's AI coding tech and hire key talent.
SpaceX Backs NVIDIA Exclusively: Musk said SpaceX will use only NVIDIA AI systems, boosting NVIDIA shares.
AI Search Boosts Shopify: AI search sends more shoppers to Shopify, reversing a trend seen across online publishing.
AMD Slides: AMD shares fell as investors sought bigger AI gains, with SpaceX's NVIDIA-only decision adding pressure.
Asian Tech Stocks Rally: SoftBank jumped 13% as Asian chip stocks climbed with Wall Street's AI rally, fueled by strong earnings.
OpenAI Settles Hiring Bias Claims: OpenAI will pay $3.2M to settle DOJ claims it favored visa holders over U.S. applicants.
POLL RESULTS: REAL OR AI
Can You Still Tell if an Image Is AI?
Yesterday's image was AI. Here's how you voted: More than half (55%) thought it was real, while 45% guessed AI correctly.
 | 🟩🟩🟩🟩🟩🟩 Real (55%) 🟨🟨🟨🟨⬜️⬜️ AI (45%) |
