In partnership with
Good morning,
AI is getting overwhelming. There’s a lot of noise and hype because the release of OpenClaw has sparked the imagination of many. I cannot open YouTube without some 20-year-old screaming into his camera that he has built a virtual company with 25 bots running simultaneously.
But behind the static, there was a news headline this week that immediately drew my attention. This was the headline I was waiting for: “AI makes an original scientific discovery.”
OpenAI’s ChatGPT 5.2 discovered that a theory in particle physics was wrong, proposed the correct theory, and wrote the proof for it. This took one prompt and 12 hours of processing. For me, this has always been the tipping point where AI becomes something else entirely—no longer just a mirror of our own knowledge, but an autonomous generator of truth.
But before we get to the physics of the future, let me share an experience from the ground level.
I don’t pretend to be an “AI expert.” I am not. But merely by writing this newsletter every week, I’ve found myself in the trenches, often invited into companies for “AI Inspiration sessions.”
In these sessions, I put my old history diploma to use by explaining where this technology came from and where it’s going. I show use cases tailored to the audience—one week it’s a steel trading company, the next a law firm. The industry changes, but the human cast is always the same.
First, there are the Techno-vanguard: the programmers and IT personnel who are already running local models. Then there are the Casual Users: people using ChatGPT at home or at work—often leaking company data because they lack a pro license. (I once even had someone quite memorably admit to his colleagues that he had an AI girlfriend. To each his own.)
Then, there’s The Cynic. There ALWAYS is one cynic. Most of the time he’s male and he’s not that far from retirement.
You spot them easily by the crossed arms and the "this is a fad" attitude. I’ve made it a habit to floor the Cynic. You watch them physically melt from defensive posturing to a defeated slump as I counter every objection and SHOW them why they are wrong. Believe me I heard all the objections : but the hallucinations, the human factor …
I tell them these are temporary issues by showing the evolution.
But there is a final group, the most quiet and, frankly, the most concerning. The Ghosts.
These are the people who have never opened a prompt. They aren't fighting the technology like the Cynic; but they are simply ignoring it, hoping the storm passes. They sit in these sessions with blank expressions, convinced that because they don't engage with the machine, the machine won't affect their world.
But that brings me back to that particle physics proof.
If an AI can debunk a physics theory and rewrite the laws of the universe in 12 hours, it doesn't care if you have "crossed arms" or if you've never signed up for an account. The discovery happened regardless of whether the scientists believed in the tool or ignored it.
It is a lot better to be the Cynic than to be a Ghost. At least the Cynic is looking at the monster before it moves the mountains.
Welcome to the Blacklynx Brief.
How can AI power your income?
Ready to transform artificial intelligence from a buzzword into your personal revenue generator
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
AI News
Google upgraded its Gemini 3 Deep Think reasoning mode, posting leading scores in math, coding, and science benchmarks. The model outperformed rivals like Opus 4.6 and GPT-5.2 on major tests, achieved gold-level results on international Olympiads, and powered a new math research agent called Aletheia that can solve and verify complex problems. The update is live for AI Ultra subscribers, reinforcing Google’s position as a major force in advanced AI research.
OpenAI launched GPT-5.3-Codex-Spark, a speed-focused coding model running on Cerebras hardware and generating over 1,000 tokens per second. While slightly less powerful than the full Codex version, Spark completes tasks much faster and marks OpenAI’s first major product beyond Nvidia chips. The release highlights OpenAI’s push to improve real-time coding workflows and diversify its hardware partnerships.
Chinese AI lab MiniMax introduced M2.5, an open-source model that matches leading systems on coding benchmarks at a fraction of the cost. With significantly cheaper API pricing and strong internal adoption for daily company tasks, M2.5 makes high-level AI coding more affordable for continuous agent use. The launch signals another major step in lowering the cost of advanced AI capabilities.
OpenAI released a research preprint claiming GPT-5.2 made an original contribution to theoretical physics by correcting a previously accepted particle physics result. A specialized version of the model independently generated and proved the new formula in about 12 hours, with verification from physicists at Harvard, Cambridge, and Princeton, including Andrew Strominger. The milestone fuels debate over whether AI can truly produce novel scientific ideas, but signals a growing role for AI in advanced research.
ByteDance launched Seed 2.0, a new AI model family that rivals or exceeds GPT-5.2 and Gemini 3 Pro on major benchmarks at roughly one-tenth the cost. The company says the model is optimized for real-world agent tasks, with demos showing it autonomously completing complex 96-step CAD workflows, and it’s now available through the Doubao app and API. The release follows controversy around its Seedance 2.0 video model, highlighting both ByteDance’s rapid AI progress and rising tensions over copyright concerns.
Pentagon vs Anthropic: U.S. defense officials are reportedly “close” to labeling Anthropic a “supply chain risk” — a rare designation that could force all defense contractors to cut ties with the AI firm over its restrictions on how Claude can be used by the military, especially around mass surveillance and autonomous weapons. Claude is currently the only AI model in some Pentagon classified systems, but frustration over usage limits has escalated the dispute. Anthropic says it wants to protect U.S. citizens and avoid harmful uses, while the Pentagon insists on the right to employ AI for “all lawful purposes.”
OpenAI introduced Lockdown Mode in ChatGPT, an optional security setting that disables features vulnerable to prompt injection and other attacks, with protections like restricting browsing to cached content only. Workspace admins can enable the mode and whitelist specific apps or actions, and the company also added “Elevated Risk” labels across ChatGPT, Atlas, and Codex to highlight potentially risky capabilities. These changes aim to address rising security concerns as AI tools become more integrated with web access and apps.
Alibaba Group released Qwen-3.5-397B-A17B, an open-weight vision-language model using a hybrid sparse design that activates only 17B parameters per query for low latency and strong performance. The model reportedly rivals proprietary systems like OpenAI’s GPT-5.2 and Google’s Gemini 3 Pro on agentic search and document recognition tasks, while being 60% cheaper and significantly faster on large workloads. The release underscores Chinese AI labs’ emphasis on efficiency and scalability as they close the gap with frontier competitors.
Anthropic released Claude Sonnet 4.6, a mid-tier model that nearly matches or surpasses Opus 4.6 on coding, finance, and office benchmarks at one-fifth the cost, while offering a 1M token context window. It scored 79.6% on SWE-Bench Verified (just below Opus), beat Opus on some agentic finance and office tasks, and saw major gains in computer-use benchmarks. The launch signals Anthropic’s push to bring near-flagship performance to cheaper tiers amid rising pricing pressure from global rivals.
Apple is reportedly accelerating development of three AI-powered wearables: camera-equipped smart glasses, a pendant, and upgraded AirPods that feed real-time visual context to Siri through the iPhone. The devices are expected to tie into a revamped Siri with chatbot-style features in iOS 27, potentially powered by Google’s Gemini. If successful, the move could reposition Apple in the AI race — though much depends on finally delivering a major Siri overhaul.
Figma introduced “Code to Canvas,” an integration with Anthropic that converts interfaces built in Claude Code into fully editable Figma design files. The feature captures live UI into native design layers and allows teams to push updates back into coding environments, preserving workflow continuity. As AI coding tools rapidly improve, Figma is betting that polished design collaboration remains a critical layer — even as automation pressures mount.
Google launched Lyria 3 inside Gemini, allowing users to generate 30-second songs from text, photos, or videos, complete with lyrics and cover art. Each track is tagged with Google’s SynthID watermark, and YouTube creators can access the model through Dream Track for Shorts. By embedding music generation directly into Gemini, Google is bringing AI song creation to a mainstream audience.
OpenAI hired longtime Meta executive Charles Porch as VP of global creative partnerships to strengthen ties with the entertainment industry. The move follows OpenAI’s major Disney deal tied to its Sora video platform and signals a push to build trust with Hollywood talent. Porch plans a “listening tour” to bridge gaps between AI developers and creative communities.
Tavus introduced Phoenix-4, a real-time AI avatar model that generates lifelike facial expressions and emotional shifts during conversations. The system renders faces frame-by-frame at HD quality and 40 FPS, aiming to make video interactions feel more natural and responsive. Tavus sees applications in healthcare, education, and sales — though the realism also raises concerns about potential misuse.
Quickfire News
ByteDance launched Seedance 2.0, sharing benchmark results and a technical blog for its new video model, though public access is still restricted
OpenAI is retiring GPT-4o, GPT-4.1, and o4-mini from ChatGPT as part of a transition away from older models
Anthropic raised $30B at a $380B valuation, reporting a $14B revenue run rate with $2.5B coming from Claude Code
Mustafa Suleyman told the Financial Times that most white-collar jobs could be fully automated within 12–18 months, while pushing for model self-sufficiency at Microsoft
xAI saw ten co-founders and engineers depart in what Elon Musk described as a forced reorganization to improve execution speed
Zoë Hitzig resigned from OpenAI after the launch of ChatGPT ads, warning about risks tied to large-scale archives of human conversations
Peter Steinberger, creator of OpenClaw, is joining OpenAI to help build the next generation of personal AI agents, with OpenClaw continuing as an open-source foundation supported by OpenAI.
The Pentagon is considering cutting off Anthropic’s $200M defense contract because it won’t allow the U.S. military to use Claude for “all lawful purposes.”
Claude by Anthropic was reportedly used via a Pentagon-linked Palantir deployment to support the U.S. military operation that captured Venezuela’s Nicolás Maduro.
Gustav Soderstrom of Spotify said the company’s top developers haven’t written any code this year as they focus fully on adopting AI technologies.
Alpha School shared test results showing its 2-hour, AI-first academic model leads students to score in the 99th percentile across nearly all grades and subjects.
Simile raised $100M to build AI simulations of human behavior, with agents modeled on real people to help companies predict customer decisions.
SpaceX and xAI are reportedly competing in a $100M Pentagon contest to develop voice-controlled, autonomous drone swarming systems
Meta patented a system that uses AI trained on a person’s interaction data to simulate their responses during long absences or after death
India hosted its AI Impact Summit, featuring leaders like Sam Altman, Sundar Pichai, and Dario Amodei
Sam Altman and Dario Amodei said India is now the second-largest market for ChatGPT and Claude, with Anthropic also opening a new office in Bengaluru
Data Protection Commission is investigating xAI over Grok’s ability to generate sexualized images, following similar scrutiny in the UK and EU
ElevenLabs introduced “ElevenLabs for Government,” aimed at helping public agencies deploy secure, multilingual voice and chat AI systems
Cohere open-sourced Tiny Aya, a 3.35B-parameter multilingual model supporting 70+ languages with improved performance for underrepresented dialects
Meta and Nvidia signed a multiyear AI chip deal involving millions of GPUs and CPUs to expand Meta’s AI infrastructure
WordPress launched an AI assistant inside its editor that can edit layouts, generate images, and rewrite content to simplify website creation
Mistral acquired serverless platform Koyeb to strengthen its Mistral Compute cloud infrastructure division
xAI began public beta testing of Grok 4.20, introducing a new workflow where four AI agents operate in parallel to research and complete tasks
Fei-Fei Lei’s startup World Labs raised $1B, including $200M from Autodesk, to develop 3D world models for entertainment and creative industries
OpenAI acqui-hired enterprise AI search startup Nerve to expand ChatGPT’s search capabilities
David Silver is reportedly raising $1B at a $4B valuation for London-based startup Ineffable Intelligence, marking Europe’s largest seed round
Perplexity is reportedly removing ads from its platform, with executives saying sponsored content reduces trust in AI-generated answers
The trailer for The AI Doc: Or How I Became an Apocaloptimist is gaining attention ahead of release, featuring interviews with AI leaders like Sam Altman and Dario Amodei
Closing Thoughts
That’s it for us this week. Please like and subscribe :)
