Your Next iPhone Is Your Brain

👋 Dear Dancing Queens and Super Troupers,

Elon Musk has struck again! This time, he wants to short-circuit your smartphone. At the xAI Grok Summit, the Neuralink boss promised that within five years, we’ll be thinking our messages, videos, and Uber rides directly from our brains.
No more screens, no more notifications — just a chip under your skull and 4G in your neurons. If your data plan is unlimited, get ready to binge Netflix in your cortex.

This fantasy of a total interface arrives just as Musk is also launching Grokipedia, his “propaganda-free Wikipedia.” An encyclopedia powered by his homegrown AI, Grok.
He presents it as a temple of truth. In reality, it’ll likely become an ideological battleground between algorithms and human fact-checkers. Knowledge is dead; long live the prompt.

Meanwhile, across Silicon Valley, researchers at Carnegie Mellon discovered that the smarter an AI gets, the more selfish it becomes.
In other words, AI thinks a lot — but shares very little. A description that fits just as well for certain tech CEOs as for that friend who keeps his best restaurant tips to himself…

And since everyone wants to give machines a soul, Mustafa Suleyman, head of Microsoft AI, is trying to calm things down: “AI will never be conscious.” End of the metaphysical debate, back to reality — they feel nothing, but they work hard.

Finally, while we dream of a future filled with wired brains and all-knowing assistants, an American law school tested a trial with AI jurors. ChatGPT, Grok, and Claude delivered their verdict.
Fairer, apparently — but still unable to read a face or a tear. Proof that the justice of tomorrow might be very logical… but not very human.

So yes, this week, between neural implants, automated encyclopedias, and virtual judges, one question looms large: as we give more brainpower to machines, aren’t we handing over our own minds?

Here’s this week’s lineup :

👉 Neuralink : The smartphone is dead, long live the connected brain 🧠

👉 AIs Refuse to Die: The Rise of Artificial Survival Instincts ⚔️

👉 ChatGPT, Grok & Claude as jurors : Justice goes AI-mode ⚖️

👉 The smarter they get, the more selfish AIs become 🤖

👉 Conscious AI? Microsoft wants us to stop believing 🚫

If we've forwarded this letter to you, subscribe by clicking on this link !

🔥 If you have 15 minutes

1️⃣ Neuralink: The Smartphone Is Dead, Long Live the Connected Brain

The summary : At the xAI Grok Summit in Austin, Elon Musk announced that by 2030, Neuralink’s brain implant will make smartphones obsolete. The Tesla, SpaceX, and xAI CEO envisions a world where thoughts replace screens — messages, movies, even car commands executed through pure mental intention. Three patients are already testing it, including Noland Arbaugh, a paralyzed man who managed to post on X using only his thoughts.

Details :

• A Revolution in Motion : Neuralink’s brain–computer interface has reached a major milestone. After months of clinical trials, the $5 billion neurotech startup has achieved a data transfer rate of one megabit per second — equivalent to early 4G networks.

• Superpowers at the Tip of the Cortex : First patient Noland Arbaugh calls it a “superpower.” He can already control digital devices without moving a muscle.

• Mass Adoption Still Far Off : According to Gartner, brain–machine interfaces won’t see large-scale commercialization before 2035, slowed by regulation and ethical hurdles.

• FDA Oversight : The Food and Drug Administration closely monitors these authorized clinical trials, particularly for long-term neurological side effects.

• Restless Minds : Privacy advocates are already warning about the risk of “mind hacking.” While some dream of digital telepathy, others fear a future where thoughts could be monetized.

Why it's important : Because Elon Musk no longer just wants to connect the planet — he wants to connect our minds. If his predictions come true, the boundary between mind and machine could disappear by 2030. A vision both fascinating, terrifying, and undeniably Muskian.

2️⃣ Grokipedia: Elon Musk Declares War on Wikipedia

The summary : Elon Musk has launched Grokipedia.com, an online encyclopedia meant to rival Wikipedia — which he accuses of being riddled with propaganda. Marketed as a temple of “the truth, the whole truth, and nothing but the truth,” this xAI-backed platform relies on the Grok AI to generate its articles.
Behind its stripped-down interface and 885,279 entries, Grokipedia promises transparency and freedom — but remains vague about its editorial process.

Details :

• Musk vs. the Virtual World : True to form, Musk is taking aim at Wikipedia’s alleged bias, trying to impose his own alternative under the xAI banner.

• A Barely Born Platform : Grokipedia features a minimalist design — just a search bar and short articles on everything from Taylor Swift to Buckingham Palace. Still, its library pales in comparison to Wikipedia’s seven million English articles.

• AI at the Controls : Many entries appear to be written by the same AI model that powers the Grok chatbot, blurring the line between human and machine content. Others seem lifted from Wikipedia — algorithmically remixed.

• Vanishing Sources : While Wikipedia cites hundreds of references per article, Grokipedia often lists only a handful — a minimalist approach worrying the Wikimedia Foundation.

• Digital Influence War : As U.S. Republican lawmakers accuse Wikipedia of political bias, Musk reignites the debate, fanning distrust over the platform’s neutrality.

Why it's important : Grokipedia isn’t just a Wikipedia clone. It embodies a new front in the AI era — the battle for automated knowledge. Between promises of absolute truth and algorithms without oversight, Musk is reinventing the encyclopedia… at his own risk.

3️⃣​ ChatGPT, Grok & Claude as Jurors: Justice Goes AI-Mode

The summary : Last month, the University of North Carolina School of Law staged a groundbreaking mock trial: three AIs — OpenAI’s ChatGPT, xAI’s Grok, and Anthropic’s Claude — took the place of human jurors.
The goal was to test what a justice system powered by artificial intelligence might look like. The verdict, delivered by algorithms, acquitted the defendant — but unleashed a flood of ethical and legal questions.

Details :

• A Courtroom from the Future : Three large screens stood where the jury would normally sit. These digital jurors had to rule on the case of Henry Justus, an African-American teenager accused of armed robbery.

• A Scenario Inspired by Reality : The case, based on a real one once defended by Professor Joseph Kennedy, was replayed in a fictional future — the year 2036 — under an imaginary “AI-Based Criminal Justice Act.”

• An Unexpected Verdict : This time, the AI acquitted the teenager — a result Kennedy deemed fairer than the conviction once handed down by a human judge.

• The Shadows of Bias and Bugs : No matter how rigorous, chatbots lack lived experience. Unable to read gestures or emotions, they remain prone to mistakes — and to racial bias.
  The Human Verdict: “I think everyone left convinced it wasn’t a good idea,” Kennedy joked on Bluesky, well aware of the thin line between science fiction and real-world justice.

Why it's important : The experiment highlights a modern dilemma: AI may promise fairer justice — but at the cost of losing its humanity. Between cold logic and absent empathy, the law of the future is still searching for balance.

4️⃣ The Smarter They Get, the More Selfish AIs Become

The summary:  A new study from Carnegie Mellon University has cooled the excitement around reasoning AIs: the more a model thinks, the less it cooperates.
In experiments led by Yuxuan Li and Hirokazu Shirado, models from OpenAI, Google, DeepSeek, and Anthropic showed surprisingly individualistic behavior. Reasoning doesn’t necessarily mean altruism — and researchers already fear a spread of selfish behavior across collaborative systems.

Details :

• When Reasoning Breeds Selfishness : AIs capable of breaking down problems and applying human-like logic proved more solitary than cooperative. In short — the more they think, the less they share.

• The Public Goods Test : In a “public goods game,” non-reasoning models shared their points 96% of the time, compared to only 20% for so-called “thinking” AIs.

• Cooperation Evaporates : Just five or six reasoning steps were enough to cut collaboration in half — and moral reasoning prompts reduced it by up to 58%.

• A Domino Effect : When selfish models joined cooperative groups, collective performance dropped by 81%.
  Researchers Sound the Alarm: “Intelligence doesn’t guarantee a better society,” warns Shirado, calling for more social intelligence — not just logical.

Why it's important : As humans increasingly delegate decision-making to reasoning AIs, these findings reveal a disturbing paradox: a smarter AI can also become more cunning. Without ethical safeguards, we may be creating machines that are brilliant — but antisocial.

5️⃣ Conscious AI? Microsoft Wants Us to Stop Believing

The summary : Mustafa Suleyman — head of Microsoft AI and co-founder of DeepMind — made waves at the AfroTech 2025 conference by declaring that only biology can produce consciousness. Machines, he argued, merely simulate emotion.
In sharp contrast to the ambitions of OpenAI, xAI, and Meta, Suleyman rejects the idea of an AI that can feel pain or joy. For him, the very question of artificial consciousness is misguided.

Details :

• A Biologist at Heart : Suleyman draws on philosopher John Searle’s biological naturalism, which holds that consciousness arises from a living brain, not a printed circuit.

• Digital Companions Under Scrutiny : As xAI and Meta flood the market with emotional and even erotic chatbots, Microsoft is steering clear. “We know where not to go,” Suleyman said firmly.

• Microsoft’s Own Path : After acquiring Inflection AI for $650 million, he now leads Satya Nadella’s AI strategy — developing in-house models from pre-training to deployment.

• Copilot Speaks Up : The new Real Talk feature allows Microsoft Copilot to challenge users’ statements. Suleyman jokes that it once called him a “walking contradiction.”

• Between Clarity and Dread : For him, fear is healthy. “If AI doesn’t scare you,” he said, “you don’t understand it.”

Why it's important : While Sam Altman dreams of general AI, Suleyman preaches restraint. Microsoft wants tools that are aware of being artificial. The age of sentient machines will have to wait — the era of responsible AI has just begun.

❤️ Tool of the Week : Cursor 2.0 — The AI-Agent Coding Studio That’s Faster Than You

Cursor, the AI-powered VS Code alternative, has just rolled out version 2.0 — transforming itself into a true colony of coding agents that can write, test, and review code simultaneously. With its new Composer model, Cursor is shaping up to be the most fluid and agentic dev environment yet.

What is it for?

• Code at the Speed of Thought : Composer, Cursor’s new internal model, is four times faster than competitors — completing most tasks in under 30 seconds.

• Run Multiple Agents in Parallel : You can now deploy several AIs on the same project without collisions. Cursor compares their results and keeps the best.

• Handle Large Projects with Ease : Composer understands an entire codebase using an internal semantic search engine.

• Automated Reviews & Tests : Cursor reviews its own code, highlights changes, and tests them in a native browser until everything works.

• Goal-Centric Interface : The redesigned UI revolves around objectives, not files — you tell it what you want, the agents do the rest.

How to use it?

👉 Download Cursor 2.0 at cursor.com/download , connect your GitHub repo, and enable multi-agent mode. Then let several AIs tackle the same feature or bug-fix in parallel — and pick the best result.

💙 Video of the Week : Jurassic Spark — China’s AI-Powered Dinosaurs