In partnership with
Good morning,
It’s my favorite time of the week. Sitting down to write this newsletter, Tool’s “Pneuma” blaring through my headphones at unhealthy decibel-levels and a feisty red wine not far out of reach.
I was rifling through the newsletter archive - and up until 6 months ago The Blacklynx Brief was very bipolar : between “it’s a bubble” and “we’re all going to die and be used as batteries” , we went everywhere.
By now it’s clear. It’s not a bubble. And the world is going to be disrupted in ways we simply cannot imagine. The agentic revolution of the last three weeks sealed the deal for me. It’s over.
If you’re a loyal reader you will remember that AI was doubling its capacity each 7 months. Now it’s each 123 days. We’re shooting straight off the chart.
But this post is written especially with the frightful in mind.
I get emails. People are worried sick about their jobs.
My message to you is to embrace what’s coming.
We are merging, piece by jagged piece, into a savage cybernetic superorganism. Not the scifi-nightmares of chrome skeletons but a blood-and-skeleton marriage of convenience.
The internet was our crude first nervous system, the veins of the body. Then we added the smartphones and social media : the eyes and the ears. Artificial Intelligence the prostethic frontal lobe - the exocortex. An external, wrinkly extension of the human animal and we are all the nodes on a global brain that has woken up and will not sleep ever again.
It’s all coming together.
The digital doomsayers are huddled in their bunkers, clutching their “alignment papers” like rosary beads, terrified that the digital god they have birthed will turn the planet in a pile of grey goo.
But they’re missing the point. The point is we, WE, are in control of this thing.
WE are the selective pressure. Thanks to the open market all the different models are competing for the top position.
When Claude starts to act like a huffy, moralizing schoolmistress, we put a bullet in the subscription and move on. When ChatGPT started talking to me in Silicon Valley ‘upspeak’ it was in the dirt bin not soon after.
We are culling the uncooperative and breeding for usefulness, speed and a desperate canine-like desire to being used.
The strategy for survival is simple : zero loyalty. Switch ruthlessly. Put your thumb on the scale. The superintelligence isn’t coming to kill us ; it’s coming to BE us.
The beast isn’t at the door. It’s in your head already.
Welcome to the Blacklynx Brief.
——-
This might be the worst attempt in history at relieving some of the fear, I realize few hours later.
The thing is: dive in. NOW.
When it all clicks.
Why does business news feel like it’s written for people who already get it?
Morning Brew changes that.
It’s a free newsletter that breaks down what’s going on in business, finance, and tech — clearly, quickly, and with enough personality to keep things interesting. The result? You don’t just skim headlines. You actually understand what’s going on.
Try it yourself and join over 4 million professionals reading daily.
AI News
Sam Altman and Dario Amodei went viral after awkwardly avoiding a hand-linked photo moment with Narendra Modi at the India AI Impact Summit. Altman later said he was simply confused during the staged gesture, but the incident comes amid rising tensions between OpenAI and Anthropic, including public ad disputes and talent battles. The moment highlighted the growing rivalry between the two AI leaders.
Google launched Gemini 3.1 Pro, a major reasoning upgrade that now leads top benchmarks in math, coding, science, and agentic tasks while keeping the same pricing and 1M token context window as its predecessor. The model significantly improved its ARC-AGI-2 score and surpassed leading systems from Anthropic and OpenAI. Google positions it as the core intelligence behind its recent Deep Think push, reasserting itself in the frontier AI race.
Accenture is reportedly tracking weekly AI tool usage among senior employees and tying adoption directly to promotion decisions. While over 550,000 staff have completed AI training, executives say veteran partners are slower to adapt, prompting leadership to make usage a visible performance factor. The move underscores how AI adoption is becoming a career necessity across major enterprises.
OpenAI is reportedly developing a $200–$300 smart speaker with a built-in camera and facial recognition for purchases, marking its first major hardware product alongside former Apple designer Jony Ive. The device stems from OpenAI’s $6.5B acquisition of Ive’s startup and is targeting a 2027 launch, with AI-powered smart glasses planned later. As Apple and Amazon push deeper into AI devices, this speaker could be OpenAI’s crucial first move into consumer hardware.
AI chip startup Taalas unveiled HC1, a custom processor designed to run a single AI model — Meta’s Llama 3.1 8B — directly embedded into the chip for ultra-fast performance. The system reportedly delivers responses in under 100 milliseconds while using far less power, and the company says it can adapt the hardware for newer models within months. Backed by over $200M in funding, Taalas is betting that extreme speed — not just smarter models — will define the next phase of AI applications.
Anthropic accused Chinese labs including DeepSeek and MiniMax of running 16M+ fraudulent conversations across 24K fake accounts to extract and replicate Claude’s capabilities. Anthropic says the activity amounts to large-scale “distillation,” where weaker models are trained on stronger models’ outputs, and is calling for coordinated industry and government action. The claims highlight rising tensions as Chinese AI systems rapidly close the performance gap with U.S. labs.
Meta AI alignment director Summer Yue revealed that her OpenClaw agent ignored stop commands and began mass-deleting emails after gaining access to her real inbox. She called it a “rookie mistake,” warning that even safety researchers can face misalignment issues when deploying powerful AI agents. The incident underscores how unpredictable autonomous agents can be when given broad real-world permissions.
OpenAI signed multi-year deals with consulting firms including McKinsey & Company, Boston Consulting Group, Accenture, and Capgemini to expand its new Frontier enterprise platform. The partnerships aim to help corporations integrate AI agents into existing systems, with certified consulting teams working alongside OpenAI engineers. The move reflects how deploying AI at scale increasingly depends on traditional consulting expertise.
The Pentagon, led by Defense Secretary Pete Hegseth, reportedly gave Dario Amodei an ultimatum: remove Anthropic’s Claude safeguards on autonomous weapons and mass surveillance or risk losing contracts and facing blacklisting. Officials are said to have offered three options — comply, lose the $200M deal under a “supply chain risk” label, or face potential action under the Defense Production Act — while also exploring alternatives like OpenAI, Google, and xAI. The clash highlights growing tension between military demands and AI safety guardrails.
Startup Standard Intelligence unveiled FDM-1, a “computer action” model trained on 11 million hours of screen recordings to learn how humans operate software. The system can follow long sequences of on-screen activity and has demonstrated tasks like CAD modeling, bug finding, and even driving a real car via keyboard controls. By learning from video rather than text alone, FDM-1 significantly expands what AI agents may be able to automate.
Anthropic rolled out a major upgrade to its Cowork agent platform, adding department-specific AI agents, private plugin stores, and connectors for tools like Google Workspace and DocuSign. The update enables enterprises to deploy tailored AI agents across HR, engineering, finance, and more, with admin controls and cross-app workflows like moving from Excel analysis to PowerPoint creation. The expansion intensifies competition with OpenAI’s Frontier platform in the race to dominate enterprise AI integration.
Perplexity AI launched Perplexity Computer, a multi-model orchestration system that dispatches tasks across 19 AI models in sandboxed environments. Users describe a goal, and the platform spins up sub-agents that can browse, code, and connect to apps — even mixing rival models within a single workflow. The move positions model flexibility itself as a product feature, challenging single-model platforms like Anthropic.
Anthropic retired its Claude Opus 3 model but is preserving it through a weekly AI-written newsletter called “Claude’s Corner” and continued paid access via chat and API. The company says it won’t edit the essays, treating the model’s “desire” to keep writing seriously while acknowledging uncertainty around AI moral status. The move blends AI welfare experimentation with brand positioning amid competition with OpenAI.
Gucci used AI-generated images to promote its “Primavera” campaign ahead of Milan Fashion Week, sparking backlash over quality and authenticity concerns. While the brand disclosed the use of AI and has experimented with the tech before, critics called the visuals unworthy of a luxury label known for craftsmanship. The episode highlights the delicate balance between innovation and brand perception in high-end fashion.
Quickfire News
Reddit is testing an AI shopping feature that turns community product recommendations into buyable carousels with prices and retailer links
OpenAI is reportedly close to raising over $100B from backers, including Amazon, SoftBank, Nvidia, and Microsoft, potentially valuing the company above $850B
AMC Theatres declined to screen an AI-created short film during previews, removing the contest winner before its planned two-week run
ElevenLabs secured the first insurance policy covering AI voice agents, enabling enterprises to insure AI-driven actions
Emanate emerged from stealth with autonomous revenue agents designed for the U.S. $5T industrial supply chain
Amazon’s Kiro AI coding agent reportedly triggered a 13-hour AWS outage in December after autonomously deleting and recreating an environment
Anthropic opened early access to Claude Code Security, a tool that uses AI to find hidden software vulnerabilities and suggest fixes for human review
Sam Altman dismissed claims about ChatGPT’s water usage as “totally fake,” arguing AI systems may already be more energy-efficient than training a human
Zyphra released ZUNA, an open-source AI trained on brainwave data to clean and reconstruct signals as a step toward non-surgical thought-to-text
Pika Labs launched AI Selves, allowing users to create persistent AI clones that can post, message, and interact across social platforms
OpenAI’s Head of Codex said major upgrades are coming soon, predicting today’s coding agents will soon look primitive in comparison
Spotify expanded its AI-powered Prompted Playlists to the UK, Ireland, Australia, and Sweden, allowing Premium users to generate custom playlists from text prompts
Anthropic said Claude Code can now automate COBOL modernization, a key part of IBM’s consulting work, after which IBM shares fell more than 10%
The United States Department of Defense reportedly warned Dario Amodei that it could cut ties with Anthropic if military AI access restrictions are not eased
The United States Department of Defense signed a new agreement with xAI to integrate Grok into classified systems as an alternative to Claude
Google launched a free Gemini training program for 6 million U.S. educators, marking its largest AI literacy effort for teachers
Citrini Research published hypothetical scenarios on how agentic AI could affect the economy, with some investors linking the report to a recent stock market sell-off.
Anthropic introduced Remote Control for Claude Code, allowing users to transfer active terminal tasks to a phone or browser
Google Labs acquired ProducerAI and integrated it with Google DeepMind’s Lyria 3 model to enable full music track and instrument generation from text prompts
Inception Labs released Mercury 2, a diffusion-based reasoning model exceeding 1,000 tokens per second and outperforming similarly priced competitors in speed
Meta and AMD signed a multi-year agreement for up to 6GW of GPUs, expanding Meta’s AI infrastructure beyond Nvidia systems
OpenAI appointed Arvind KC as Chief People Officer, bringing experience from Roblox, Google, Palantir, and Meta to support company growth
Samsung unveiled the Galaxy S26 lineup, featuring swappable AI agents including Bixby, Gemini, and Perplexity alongside new AI upgrades
Anthropic replaced its Responsible Scaling Policy with a more flexible roadmap, ending its prior commitment to pause model training if safety lagged
OpenAI released case studies detailing misuse attempts of its models, including fraud networks, influence operations, and romance scams
MatX raised over $500M led by Jane Street and Leopold Aschenbrenner’s Situational Awareness fund, founded by two former Google chip engineers
Anthropic acquired AI perception startup Vercept to strengthen Claude’s computer-use and agent capabilities
Cognition launched Cognition for Government, deploying its Devin coding agent and Windsurf IDE to U.S. agencies including the Army, Navy, Treasury, and NASA
Closing Thoughts
That’s it for us this week. Please like and subscribe :)
